Featured Video Play Icon

Do we reflect while performing skillful actions? Automaticity, control, and the perils of distraction

Juan Pablo Bermúdez
 (Universidad Externado de Colombia)

[Jump to comments]


Abstract: Many human actions require skill, from our everyday commuting to the gold medalist’s world-class performance. What is the proper explanation of skill-involving action? There are two main answers: anti-intellectualism holds that skillful action is produced by automatic coping processes without the involvement of higher-order mental processes; intellectualism claims that even bodily skillful actions require the involvement of intention-directed processes. I examine the evidence for three psychological phenomena recently held to support anti-intellectualism—choking under pressure, expertise-induced amnesia, and expert confabulation—, and argue that, while the evidence contradicts ‘Anscombean’ kinds of intellectualism, it also supports another intellectualist view, namely that the executive, top-down control of attention is a necessary component of any skilled action.

Keywords: Skill, expertise, habit, attention, automaticity, control, utilization behaviour, dual process.


 1. Introduction

Actions involving skill and expertise are particularly controversial. The remarks of Niki Nakayama, an expert chef, about the way she works, illustrate this:

Cooking is the one thing that I feel I can completely trust what I’m doing. When I’m planning a dish, my mind is completely shut off. It’s all based on feeling: ‘this has to be here, this has to be here, this feels right here, this looks right here.’ I think it’s similar to that meditative state that people can get to, where they’re not listening to their minds anymore, but it’s just that moment.

The claim that her mind is not involved when acting skillfully seems to go against a broadly shared view (which I will call intellectualism) according to which action production requires high-order mental processes involving some state like intentions, beliefs, or desires.[1] Her description rather resembles Csikszentmihalyi’s description of flow, a state in which “there is no need to reflect, because the action carries us forward as if by magic” (1990, 155). But notice another of Nakayama’s remarks:

The best advice that I’ve been given was to never stop learning. Because the moment you give up and think you know everything, you’re done as a chef.[2]

Why should an expert keep learning, if her mind is shut off during performance? The coarse-grained reflection required for learning something new or altering overpracticed actions may hinder intuitive flow. So if experts are continuously learning, this suggests that the mind is more closely involved in skillful action than previously suggested.

It has been recently argued that automatic processes have a norm-sensitivity of their own, based on resolving felt tensions.[3] Assuming this is correct, is automatic control sufficient for expert action, or must it include some element of reflection? Anti-intellectualists defend the former view: the more skilled you are, the less you need to think about what to do or how to do it; the practical situation simply solicits the proper response from you. As in Daoist wei-wu-wei,[4] the expert ‘acts without doing’: responds without planning or calculating by merely resonating with her practical environment’s dynamic lines of force. Intellectualists take the latter view, often pointing to the constant drive for improvement as evidence of reflection’s centrality to expertise. This drive, they argue, is what separates true experts from merely competent performers: complete automaticity begets stagnation, so experts keep a window, however small, for reflection to continue directing and structuring automated routines, thus increasing performance control.[5] Nakayama’s claim that while cooking her action is based solely on feeling supports anti-intellectualism; her further statement that a chef must never stop learning supports intellectualism. So what kind of control do skilled and expert agents exert while performing skillful actions?

Intellectualists face a tough task, since the fast, fine-grained, adaptive bodily control displayed in skillful performance seems hardly explainable by such coarse and slow phenomena as deliberation and planning. Moreover, intellectualists also must explain how reflective elements are integrated with the automatic processes that control the largely automatic routines of skilled action. Intellectualists have recently taken up the task, arguing that appeals to sub-personal, automatic processes leave agentive control, the crucial feature of skilled action, unexplained.[6] Some anti-intellectualists have recently replied to this challenge by pointing to a number of psychological phenomena that they claim intellectualism could not explain, because they show that reflecting on what one is doing is either unhelpful or even harmful to the performance of skillful actions.[7]

This essay argues that said evidence, if examined closely, reveals that reflection does play a necessary role in skilled action. Automatic control is insufficient for skilled action because it cannot explain the coherence and unity between automatic routines that any skillful action requires. Before I present my argument for these claims, allow me to introduce the three empirical phenomena recently cited in support of anti-intellectualism.

1.1. Three psychological phenomena against intellectualism


Choking / the yips

Some intellectualists hold that automatic behaviour counts as intentional action only if the agent can answer ‘Anscombean’ questions (like ‘What did you do?’ and ‘Why did you do that?’) about it.[8] Against this kind of intellectualism, Michael Brownstein has invoked a phenomenon variously called “choking”, “the yips”, and “Steve Blass disease”:

Steve Blass was a hugely successful Major League Baseball player who inexplicably lost his ability to pitch. […] Suddenly and inexplicably, he couldn’t even hit the catcher’s mitt from the mound. […] The common interpretation is that Blass’ problem stemmed from something like chronic over-thinking. This is a common enough ‘‘disease’’ in profession sports to earn its own name. (Brownstein 2014, 557)[9]

Brownstein argues that the reflective self-awareness necessary to answer Anscombean questions is precisely the kind of over-thinking that causes yipping. Multiple studies seem to confirm Brownstein’s claim,[10] which in turn supports the anti-intellectualist view that the fine-grained sensorimotor control of expert performance does not involve reflection.


Expertise-induced amnesia

The automation necessary to acquire a skill seems to generate expertise-induced amnesia,[11] a disconnection between what is done and what is accessible to self-report. Many expert testimonies corroborate this. Take basketball star Larry Bird: ‘‘[a lot of the] things I do on the court are just reactions to situations […] A lot of times, I’ve passed the basketball and not realized I’ve passed it until a moment or so later’’; Hall of Fame NFL running back Walter Payton: ‘‘people ask me about this move or that move, but I don’t know why I did something. I just did it’’; and philosophical baseball catcher Yogi Berra: “Think? How can you hit and think at the same time?”[12]

Expertise-induced amnesia speaks against intellectualism about expert action. If an agent’s behaviour implies a reflective process, some aspects of what she did and how she did it must have been accessed by working memory; and if so, the agent should be able to report them (since reportability and working memory seem to be closely linked).[13] The unreportability of control processes suggests that skilled actions are controlled automatically.


Expert confabulation

Expertise-induced amnesia notwithstanding, some experts do produce reports about what they do and how they do it. But such reports often seem to be mistaken. Studies suggest that the rules experts believe to be following (about e.g. how and where to focus attention) turn out to misdescribe what they actually do. Cricket batters, for instance, are taught to always keep their eyes on the ball, and experts seem to believe they do. But eye-tracking evidence suggests that the more experienced a batter is, the less time her eyes spend focusing on the ball: the expert batter’s eyes move quickly away from the ball, making a “predictive saccade” to the place where they anticipate the ball will bounce.[14] The more expert the player, the faster the saccade; so expertise is inversely proportional to the time spent watching the ball.

Cases like this (and, as we will see, this is not the only one) seem to support the anti-intellectualist view that skill is inversely proportional to reliance on explicit rules. Moreover, the fact that the performer’s explicitly held rules and beliefs misdescribe performance suggests that the semantic content of her reflective states and intentions does not (or at least not entirely) control skillful automatic processes.


Do these three psychological phenomena provide solid support for anti-intellectualism about skilled action? Before moving toward an answer, allow me first to clear a few conceptual issues.

1.2. On the notion of ‘skill’


Skill and automaticity

The precise meaning of ‘skill’ is part of what is at stake here, but it seems clear that having a skill implies having automated the performance of some cognitive or sensorimotor sub-routines through practice. Thus, skillfully doing A implies that certain components of A-ing have become automated, or ‘chunked’, and are thus performed without recruiting working memory. (E.g. skillfully typing these words implies having automated the processes that associate specific letter strings to specific finger movements, so that I no longer reflect about finger movements while writing each word).

Automaticity is thus necessary for skill. But it does not seem to be sufficient, because not all acquired associations would qualify as skills (e.g. those that constitute implicit biases, or mere associations between ideas). What else does skill need, then?


Skill and control

In order to count as a skill, and not merely a collection of automated processes, at least an element of control is required that unites automatic routines into a coherent performance.[15] Skilled agents are able to fine-tune each of their body movements to the particular situation, direct their attention to exactly the relevant features, and make multiple precise corrections on the fly.

Anti-intellectualists hold that automatic control, along with the agent’s history of long practice and habituation, can account for such fine-grained, fast-paced control, and that online reflection about these processes, being too effortful and coarse-grained, would lead to the loss of control. Intellectualists hold that automatic control cannot sufficiently explain the broader coherence of skillful performance, or the precision of world-class, expert performance. The task ahead is establishing which view is more strongly supported by evidence and argument.


Kinds of skills

‘Skill’ and ‘skilled action’ are quite ambiguous terms. It seems unlikely that the skills involved in my tying my shoelaces, a cricket batter’s batting, a physician’s reaching a complex diagnosis, and a philosopher’s critically assessing an argument, all share some common essence. I will thus focus here on sensorimotor skills involved in bodily performances, like sports and dance, which seem like more of a unified class, and are the ones most traditionally problematic for intellectualism.


Skill and expertise

Some researchers understand skill as a continuum between novices and experts, and thus conceive of expertise as nothing but the highest degree of skill.[16] Others think skill and expertise are qualitatively different, general skill largely consisting in automatic dispositions (and therefore susceptible to stagnation), expertise including online reflective guidance toward continuous improvement.[17] Intellectualists tend to support a qualitative distinction, whereas anti-intellectualists lean towards a quantitative distinction. It is important not to presuppose either view, rather assessing the arguments for and against each.


2. Reflection and skilled performance: Interpreting the evidence

2.1. Choking, The Yips, and the focus of attention

According to Brownstein, choking, or the yips, reveals that “performing skilled actions well stands in an inverse relationship to self-focused awareness” (2014, 557). A performer’s reflecting on her ongoing behaviour—in a way that would allow her to answer Anscombean questions—hinders skillful performance. But let us take a step back: it is unclear whether ‘choking’ and ‘the yips’ are the same thing, different phenomena with common underlying causes, or independent conditions. Moreover, if the phenomena have to do with reflective over-thinking, it is unclear what it is that people are reflectively thinking about when they suffer from these problems, and whether the effects occur when people reflect on some features but not when on others. So more clarity is needed.


2.1.1. Clarifying the terms

On the eve of the 1994 soccer World Cup, the Colombian national team was among the competition’s favourites: they had a brilliant set of players at the peak of their careers, and showed top-level performances in the qualifying games, famously beating Argentina, another favourite, 5–0. But when the moment of truth came, they failed. Actually, ‘failed’ does not do it justice. They flopped horrendously, going back home after group-stage elimination. This was a clear case of choking under pressure, in the psychologists’ sense of the term, i.e. a performance that is below what is expected given the agent’s skill level, in a high-pressure situation (i.e. a situation where the agent feels that performing well is highly important).[18]

There are two main theories about why skilled performers choke under pressure.[19] Explicit monitoring theory holds that high pressure leads the agent to closely and reflectively monitor previously automated aspects of performance, thus disrupting automatic control and leading to performance breakdown. Distraction theory states that high pressure causes the agent to reflect on aspects extraneous to her task (e.g. the outcomes at stake), which thus occupy attentional and working-memory resources, thereby disrupting performance. Each theory points to a different relation between reflection and choking: the former suggests that reflective focus on the task leads to choking; the latter, that reflective focus on the task is required to avoid choking. So which one is right?

Both are, apparently. Two different kinds of pressure seem to have two different effects on reflection, which in turn seem to cause two different performance failures. DeCaro et al. (2011) showed this by distinguishing between two kinds of pressure—the pressure of being monitored by others, and the pressure of high-stakes, performance-dependent outcomes—, and distinguishing between two kinds of tasks—reflective tasks (which rely heavily on working memory), and intuitive tasks (which do not)—. They found that the pressure of being monitored tends to hinder intuitive task performance, but not reflective task performance, whereas outcome-related pressure has the inverse effect, disrupting reflective task performance and not intuitive task performance. This suggests that the pressure of being watched can lead agents to reflectively focus on automated aspects of performance, thus hindering automatic control (as explicit-monitoring theory would suggest), while outcome pressure can lead agents to take reflective focus away from their task, thus also hindering performance (as distraction theory would hold). Of course, both kinds of pressure can coexist, like when you are playing at the World Cup—and in those cases both kinds of performance failure can occur simultaneously.

This finding suggests that ‘choking’ actually refers to at least two distinct phenomena. Papineau (2014) helpfully suggests that, although both phenomena are instances of the general phenomenon he calls “Not Having Your Mind Right”, they deserve distinct names and explanations. I follow him in using “The Yips” for cases of performance failure due to reflecting on automated components of an action, and “Choking” for cases of performance failure due to diverting reflective focus away from an action.

Also illuminating is Papineau’s (2014) distinction between basic actions (i.e. behaviours that you can directly decide to perform without having to decide to do anything else), and the components of basic actions (i.e. simpler behavioural routines that are constitutive parts of a basic action and have become automated through practice). For it seems to follow from the above that reflectively focusing on the action’s components harms performance, whereas not reflectively focusing on the basic action also does. And given this, it seems consistent with the evidence to claim that skilled agents must keep reflection away from action components, but focused on their intended basic actions. For reflecting on action components leads to loss of fine-grained automatic control, and ultimately to Yipping, while losing reflective focus on the basic action leads to distraction, and ultimately to Choking.

When action components are brought into reflective focus, fast and fine-grained automatic control is interrupted by slow, coarse-grained reflective control. This seems consistent with famous yipping cases, like those of Chuck Knoblauch, a New York Yankees second baseman who became unable to throw to first base, or Mackey Sasser, a catcher who lost his ability to throw the ball back to the pitcher, both usually interpreted as paradigmatic cases of Steve Blass disease (Blass himself became unable to hit the catcher’s mitt from the mound). They all have in common a loss of control over tasks that are relatively simple, and initiated by them. The Yips affects performers whose function is initiating plays (e.g. pitchers and catchers in baseball or bowlers in cricket), but the condition is notoriously absent from responsive aspects of performance (like batting in baseball and receiving in tennis) and continuous action sequences (like most of soccer). One can see how the pressure of being watched may may lead someone to Yipping: knowing others are waiting for me to make my move, I may worry about getting the specific performance parameters right, which would lead me to reflectively check them just before I execute them.

Choking is notably different. The cause of losing focus on the task at hand seems to be a failure to inhibit task-irrelevant features of the practical situation (like worries about what will happen if I fail), leading to the relevant features being ‘lost in the noise’. While one may advise a Yipping-prone performer to let herself go and not think too much, proper advise to avoid Choking would be to focus on the performance, to keep here eyes on the ball and her head in the game.


2.1.2. Do these phenomena support anti-intellectualism about skill?

This analysis of The Yips, Choking, and their causes, does not seem to support anti-intellectualism. The Yips does moderate intellectualism, since it shows that performers should avoid reflectively controlling action components while performing (although they may do so while practicing, e.g. while trying to correct or recalibrate a given gesture). But Choking supports intellectualism by suggesting that automatic control of action components, although necessary, must be subordinated to higher-order reflective processes of attention guidance.

In this respect, Choking is analogous to everyday slips, like when I leave the office intending to drive to the candy store, but end up driving home instead. In slips, my behaviour displays automatic control, no doubt, but the basic action’s overall unity dissolves because I fail to keep my governing intention in reflective focus, thus losing the reflective control that unifies multiple threads of performance into the basic action I intended to perform. Analogously, when performers lose focus on their overarching intention and strategy—perhaps because menacing and stressful thoughts overload inhibitory control—, this leads to performance failure. Such lack of focus can make the unity that characterizes expert skilled action dissolve into a pattern of incoherent, albeit controlled, automatic processes.

So reflective focus on the intended action seems necessary for skilled action, and the more pressure there is on the agent to perform, the more reflective control seems to be needed. Simply letting automaticity take control seems insufficient for skilled intentional action performance at the highest levels, mainly because expert-level performance situations are often crowded with distracting forces (like fans yelling things at you, fans of your team expecting so much from you, your own worries about ability or future outcomes, multiple bodily pains, sheer nervousness, etc.).

Now, what about everyday, familiar, habitual skilled actions? I would probably need to reflectively control attention in order to win the World Championship of Knot Tying, but do I need reflective control of attention in order to tie my shoelaces? The more habitual and routine the skilled action is, the less clear it is that reflection must be involved. I believe the same argument from distraction applies to both expert and habitual actions, but for reasons of space I focus here solely on expert action, and leave habitual actions for another time.


2.2. Expertise-induced amnesia

2.2.1. The anti-intellectualist case

Anti-intellectualists hold that the automation of routines implied in skill acquisition causes skilled agents to suffer from “expertise-induced amnesia” with respect to how they perform their actions. Since anything accessible to working memory is presumably available for subsequent declarative report, the failure of experts to report what they do or how they do it suggests that the cognitive processes that determine their performance must have occurred independently of working memory.[20] And given the link between working memory and reflection, the lack of reportability suggests skilled action is controlled automatically rather than reflectively.

This has suggestive empirical support. Beilock et al. (2002) studied the performance of novice and expert golfers while putting, and their reportability. Participants were asked first to list all the elements of a typical putt, and later to list all the specific steps they themselves had taken in performing their last putt. While experts gave overall longer generic descriptions of the steps involved in putting than novices, the experts’ generic descriptions were much longer and more precise than their episodic descriptions of their own putt. In fact, the shortest of all episodic descriptions were those of experts. This supports the existence of expertise-induced amnesia by showing that, although experts have a large store of general explicit knowledge about their skill, what exactly they do while performing is less accessible for them than for novices.

Intellectualists have remained unconvinced by both the argument from expert anecdotal reports, and the argument from funny-putter-type studies.


2.2.2. Intellectualist replies

It is worth noting, first, that Beilock’s studies do seem to speak against Anscombean intellectualism, by showing that more highly skilled performers seem less able to reflectively access information about the what and how of skilled performance (as measured by their ability to provide a list of the steps taken to perform the skilled action) than novices. That said, this is not sufficient to reject other intellectualist accounts of reflection’s role in skilled action. An alternative role for reflection is sustaining attention on the current task so that relevant environmental features may become salient, while irrelevant stimuli are inhibited. Expertise-induced amnesia seems insufficient to undermine this kind of intellectualism.

A further problem with taking Beilock’s experimental findings as evidence for the nature of expert action concerns the very definition of ‘expertise’. Beilock’s studies assume a gradual conception of the skill-expertise distinction, calling ‘experts’ those participants with higher degrees of skill.[21] This begs the important question whether experts are just highly skilled agents, or whether expertise is qualitatively different from skill.

Barbara Montero defends the latter. She conceives experts as “those who have practiced their skill in a thoughtful, critical manner on a near daily basis for at least around 10 years and are still practicing their s[k]ills in such a way” (2015, 386). Understood this way, Beilock’s experiments do not reveal the nature of expertise, because none of its participants are real experts. We still do not know whether professional experts would be able to produce detailed descriptions the steps they took to produce their putt or not if they participated in Beilock’s experiment.

Now, the argument from expert reports like the ones mentioned above, being merely anecdotal, is not very solid on its own. But it looks even more fragile when contrasted with other reports, like those found in studies that use retrospective methods (like diaries and verbal recollection) to find out what professional athletes think about while performing at their top level. When asked how they coped with pressure, it was found that “the most effective coping strategies that were used on a frequent basis were increasing concentration on task and increasing effort”;[22] and to maintain high-quality performance in high-pressure situations, athletes report relying on “positive monitoring” strategies (such as “I focus on what I should do” and “I make sure that I work harder”).[23]

Such reports seem to counter those in which experts claim not to know the whys or hows of their performances. It is likely that experts are unaware of why and how they structure their action components the way that they do (such structuring process consisting largely in automatic-associative responses to the practical environment’s solicitations), but one thing experts seem aware of is that to perform consistently in high-pressure situations they must stay focused on the performance itself (which implies inhibiting negative, task-external thoughts) and effortfully engage in the action. That experts describe skilled performance as effortful both physically and mentally suggests that expert action requires reflective, executive control of attention. Experts may be unaware of why and how they structure their movements during performance; but they do seem aware that in order to be able to do it, they must keep their focus on the task.

From this perspective, intellectualists can counter the anti-intellectualist interpretation of the statements of Larry Bird and others. Such statements need not be taken as evidence for the view that skilled action consists in merely automatic responses. Bird’s claim that many things he does in the court are but reactions to the situation may be taken to refer to his performance of action components rather than actions themselves; Payton’s statement that he does not know he does what he does is compatible with attention-control intellectualism; and in claiming that you cannot think and hit the ball at the same time, Berra may mean that you cannot hit properly when thinking about action sub-components or extraneous situational features—rather, you must focus on hitting itself.


2.3. Expert confabulation

2.3.1. Evidence for expert confabulation

Not all experts remain silent when asked what they do and how they do it. But when describing what they do in performance, they often mention rules they claim to follow, some of which have been found to misdescribe what they actually do. Some of the strongest evidence for such expert confabulation comes from the aforementioned study of cricket batters, according to which the popular ‘Keep your eyes on the ball’ rule—endorsed by many batters and coaches—does not coincide with what skilled performers do. Instead of following the ball, highly skilled performers perform predictive saccades to move the gaze ahead of the target, or catch up with it.[24]

Additionally, cricket training makes special emphasis on the importance of watching the ball at the precise time of hitting it. This is particularly unlikely, given the high speed at which the ball travels. Batters have been found to track the ball during the first 0.2 seconds after it is released, leave it behind to make the predictive saccade, then loosely follow it as it approaches the bat, but ultimately lose it from sight near the end of its trajectory.[25] Furthermore, predictive saccades make functional sense, since they help the batter predict the ball’s trajectory, thereby increasing likelihood of contact. Thus, experts seem wrong in reporting that they keep their eyes on the ball, that they do so at the crucial time of striking it, and even that dong it is what should be done.


2.3.2. “Expert”? “Confabulation”?

That said, the Land & McLeod study seems susceptible to the same kind of terminological objection as those of Beilock and colleagues: ‘experts’ are roughly understood as ‘those participants with the highest degree of skill’.[26] This assumes a merely quantitative difference between skill and expertise. Would true experts, i.e. world-class performers, reveal the same performance pattern?

Mann et al. (2013) tested this by comparing elite performers (players with some of the highest historical records in the game) with highly-skilled club-level performers (who play at local cricket clubs). They found a couple of relevant traits proper to elite performers. First, they consistently follow the ball throughout the whole trajectory with their heads, even if not with their eyes. Second, they make not one, but two predictive saccades: first toward the place where the ball may bounce, then toward the place where the ball may make contact with the bat. Crucially, the second saccade enables elite performers to see the ball as they hit it, much more reliably than club-level performers. Mann and colleagues compare elite batters’ capacity to follow the ball with their heads to a “miner’s torch” (2013, 6): imagining they had one, the light from the their torch would illuminate the ball throughout its whole trajectory. This ability seems to give elite players an advantage: it simplifies the cognitive problem of establishing the ball’s trajectory, since the ball’s position remains constant in egocentric space (relative to the head); so, since they already know where (in egocentric space) the ball will be, all they need to figure out is when the ball will reach them.

The ‘when’ is then specified by the second predictive saccade. In fact, a crucial difference between the two groups in the study is that elite batters moved their eyes so that they were always looking at the ball or ahead of the ball, whereas the eyes of club-level batters sometimes lagged behind, and had to make saccades to catch up with it. So only elite saccades truly deserve the name ‘predictive’: they anticipate the key moments and places of the ball’s trajectory.

Land and McLeod had concluded that skilled batters do not keep their eyes on the ball and fail to see the ball as it was hit, and that there is no “systematic difference” in eye movement between skill levels (2000, 3142), all of which would support expert confabulation. But the elite batters study puts it all into question. For one, it reveals that explicit coaching advice and performance rules are rather consistent with actual practice, since truly expert batters do watch the ball as they hit it, and they do keep their eyes on the ball throughout its trajectory, in the broad sense that the ball is always in the visual field (so it is at least peripherally visible at any time). If other cases of bodily expertise are similar, reported rules and performance may be much more consistent than previously thought.

Additionally, the study reveals a qualitative difference between real experts and merely skilled agents: experts make two saccades, whereas skilled players make only one, and only experts reliably watch the ball as they hit it. It is not just that experts are better at attending; they actually do different things with their attention.

In sum, there are reasons to doubt that the expert confabulation evidence proves the existence of confabulation in real experts.


 3. Reflection’s role in skilled action

After assessing the phenomena relative to Choking and The Yips, expertise-induced amnesia, and expert confabulation, the support of anti-intellectualism does not seem so solid. In fact, these phenomena have revealed that reflection’s involvement may be crucial in acting skillfully. This section offers a general argument for the view. I finish by considering whether there is something about skilled bodily action that anti-intellectualism cannot account for.

3.1. Reflection as the intention-directed modulation of attention

Evidence shows that while reflecting on the automated components of an action is harmful to performance (leading to The Yips), allowing reflection to drift away from the basic action itself also harms performance (leading to Choking). This suggests that reflective focus on basic actions may play a key role in unifying the different components of a basic skilled action. Further, although experts often claim ignoring why or how they do what they do, they seem to be clear about the fact that in order to do it they must make efforts to stay focused on their task, especially when under pressure. This further reinforces the view that reflection plays an attention-structuring role in the performance of skilled action. Finally, evidence from world-class experts does not support expert confabulation, rather revealing that in elite performers the general attention-orientation pattern largely coincides with explicit reports, and that differences in skill levels correlate with qualitative differences in ways of structuring attention.

From this reconsideration of psychological evidence emerges the view that top-down, intention-directed, reflective mental processes do play a necessary role in the production of skilled action, namely, the executive modulation of attention.[27]

But why would this executive, top-down modulation be necessary? An anti-intellectualist might argue that this amounts to adding an otiose element to the explanation, since past training, automated routines and the normativity of automaticity can do all the explanatory work: skillful agentive control is no more than properly trained, norm-sensitive automatic behaviour.

To see why reflection (in the form of the executive control of attention) is necessary to fully account for skilled action performance, consider this example. When playing soccer with my good friend Lucho, a great number of the situation’s aspects could activate different automatic-associative behaviours, since my practical context is crowded with affordances (which I understand here roughly and intuitively as possibilities and invitations for action). Seeing Lucho in the field affords reaching and chatting with him. But he happens to be in the opposite team, and has just scored a goal, so noticing him around also elicits frustration and affords punching him. The acute soreness in my left foot affords taking the shoe off and staying still. And while I’m at it, the warm, sunny weather may also afford tumbling on the soft grass. My soccer skills (assuming I have any) also afford a set of potential moves conducive to a good performance.

Of all these possible sensory, affective, and motor affordances—all of which are part of my possible behavioural repertoire given my past habituation and automatic control—, how can I distinguish what is relevant from what is irrelevant in the moment? The crucial lesson so far is that achieving an appropriate distinction between relevant and irrelevant affordances is hardly possible, if at all, without the top-down modulation of attention on the basis of an occurrent goal or intention. I can focus on the soccer-based affordances only if I manage to inhibit the chatting, punching, resting, etc. affordances; and doing so implies reflective, inhibitory control. And as it turns out, I can perform this inhibition only as long as I attend to the representation of my current goal. Only through such top-down control of attention can the relevant features of the situation acquire the special glow, the mandatory character that “carries me forward as if by magic” (to use Csikszentmihalyi’s expression), or that “summons” the correct response from me (to use Dreyfus’ term). If attention is not hierarchically structured by this goal, all affordances would remain equally attractive, which entails that no affordance would really be all that attractive.[28] Moreover, when the pressure is higher the task-irrelevant features have a greater intuitive salience, so inhibitory control (a limited capacity, like all executive functions) has a greater role to play.

It is because of that that anti-intellectualist descriptions of skilled action are incomplete. Take Merleau-Ponty’s classical description of a player’s experience of the field:

For the player in action the football field […] is pervaded with lines of force (the ‘yard lines’; those which demarcate the ‘penalty area’) and articulated in sectors (for example, the ‘openings’ between the adversaries) which call for a certain mode of action and which initiate and guide the action as if the player were unaware of it. […] The player becomes one with [the field] and feels the direction of the ‘goal’, for example, just as immediately as the vertical and the horizontal planes of his own body. (1942/1965, 168–169)

This description does away with reflection’s role in skillful action by mentioning only the task-relevant lines of force, and remaining silent about any others. But the field is usually pervaded with way too many lines of force, calling for way too many responses that could guide behaviour in all sorts of incoherent directions. Automatic control over the action’s subroutines is necessary to explain why the fields of force guide the action as if the player were unaware. But automatic control seems unable to explain why only the relevant lines of force guide skilled action, while action remains unaffected by irrelevant lines of force. The need to inhibit such lines of force, and select only those that are task-relevant, is the reason behind why experts recognize they have to effortfully direct attention to what they are doing. (They might otherwise explode in anger, take an untimely pause, go yell at some fans of the opposite team, have an emotional meltdown, etc.) The aspect of skilled action that remains unexplained by anti-intellectualism is the capacity to perform such inhibitory selection in the face of a myriad performance pressures, such that action components actually add up to a basic action instead of a bunch of disconnected behaviours. Prior practice and the normativity of automaticity can explain the subtleties and refinement of expert action components, but not the coherence of expert basic actions.

Thus skillful action needs reflection because orchestrating multiple automatic processes (sensory, cognitive, and motor) toward a unified action requires the modulation of attention. In this context, Wu’s (2011) suggestion that intentions are not mental events that play their causal role before action is produced (like a standard causal model would have it), but rather are structural causes synchronic with action, is quite useful. Intentions (understood broadly as representations [propositional or nonpropositional, conceptual or nonconceptual] of a goal to which the agent is committed) structure the agent’s web of possible mental associations in such a way that the relevant sensory, affective, and motor associations are primed to be more easily activated, while the activation value of irrelevant associations is decreased. Thus, by modulating the activation levels of automatic association patterns, the agent’s focus on her intention structures attention so that the relevant sensory features become salient, and hooked up with the relevant affective and motor routines. This is how I avoid hugging or punching Lucho while in the field (after the match I may do both), and instead anticipate his motions, steal the ball from him, and make a crucial pass to one of my teammates.

What emerges from the above discussion is a form of intellectualism not based on propositional knowledge of the reasons and the ability to answer Anscombean questions, but on the capacity to structure attention in relation to an intention. This intellectualist account, which we can call attentional-control intellectualism,  can make Nakayama’s claims consistent: while performing, the expert’s mind is silent about how to perform and why—but it keeps attention focused on performance. This control of attention allows her to learn continuously; indeed continuous learning may be a necessary consequence of the constant focus of attention on the task at hand, in the sense that the refinement of attention is likely to lead to discovering novel nuances and action possibilities.


Works cited

Annas, J. (2011). Intelligent virtue. Oxford University Press.

Baddeley, A. (2007). Working memory, thought, and action. Oxford University Press.

Bahill, A. T. & LaRitz, T. (1984). “Why can’t batters keep their eyes on the ball?” American Scientist 72:3, 249–253.

Baumeister, R. F. (1984). “Choking under pressure: self-consciousness and paradoxical effects of incentives on skillful performance.” Journal of Personality and Social Psychology 46:3, 610.

Beilock, S. L., Bertenthal, B. I., McCoy, A. M., & Carr, T. H. (2004). “Haste does not always make waste: Expertise, direction of attention, and speed versus accuracy in performing sensorimotor skills.” Psychonomic Bulletin & Review 11:2, 373–379.

Beilock, S. L. & Carr, T. H. (2001). “On the fragility of skilled performance: What governs choking under pressure?.” Journal of Experimental Psychology: General 130:4, 701.

Beilock, S. L. & Gray, R. (2007). “Why do athletes choke under pressure?” In Handbook of sport psychology (ed. Tenenbaum, G. & Eklund, R. C.). Wiley, 425–444.

Brownstein, M. (2014). “Rationalizing flow: Agency in skilled unreflective action.” Philosophical Studies 168:2, 545-568.

Brownstein, M. & Madva, A. (2012). “The normativity of automaticity.” Mind & Language 27:4, 410–434.

Croft, J., Button, C., & Dicks, M. (2010). “Visual strategies of sub-elite cricket batsmen in response to different ball velocities.” Human Movement Science 29:5, 751–763.

Csikszentmihalyi, M. (1990). Flow: The psychology of optimal experience. Harper & Row.

Davidson, D. (1963). “Actions, reasons, and causes.” The Journal of Philosophy 60:23, 685–700.

DeCaro, M. S., Thomas, R. D., Albert, N. B., & Beilock, S. L. (2011). “Choking under pressure: Multiple routes to skill failure.” Journal of Experimental Psychology: General 140:3, 390.

Dreyfus, H. L. (2002). “Intelligence without representation—Merleau-Ponty’s critique of mental representation. The relevance of phenomenology to scientific explanation.” Phenomenology and the Cognitive Sciences 1:4, 367–383.

––– (2012). “The mystery of the background qua background”. In Knowing without thinking: Mind, action, cognition and the phenomenon of the background (ed. Radman, Z.). Palgrave Macmillan, 1–10.

Dreyfus, H. & Kelly, S. D. (2007). “Heterophenomenology: Heavy-handed sleight-of-hand.” Phenomenology and the Cognitive Sciences 6:1-2, 45–55.

Evans, J. S. B. (2008). “Dual-processing accounts of reasoning, judgment, and social cognition.” Annual Review of Psychology 59, 255–278.

––– (2010). “Intuition and reasoning: A dual-process perspective.” Psychological Inquiry 21:4, 313–326.

Evans, J. S. B. & Stanovich, K. E. (2013). “Dual-process theories of higher cognition: Advancing the debate.” Perspectives on Psychological Science 8:3, 223–241.

Flegal, K. E. & Anderson, M. C. (2008). “Overthinking skilled motor performance: Or why those who teach can’t do.” Psychonomic Bulletin & Review 15:5, 927–932.

Fridland, E. (2014). “They’ve lost control: reflections on skill.” Synthese 191:12, 2729–2750.

––– (2015). “Automatically minded.” Synthese (published online), 1–27.

Land, M. F. & McLeod, P. (2000). “From eye movements to actions: how batsmen hit the ball.” Nature Neuroscience 3:12, 1340–1345.

Logan, G. D. (1985). “Skill and automaticity: Relations, implications, and future directions.” Canadian Journal of Psychology/Revue canadienne de psychologie 39:2, 367.

Merleau-Ponty, M. (1965/1942). The structure of behavior. Duquesne University Press.

Montero, B. (2010). “Does bodily awareness interfere with highly skilled movement?” Inquiry 53:2, 105–122.

––– (2015). “Is monitoring one’s actions causally relevant to choking under pressure?” Phenomenology and the Cognitive Sciences 14, 379–395.

Munakata, Y., Herd, S. A., Chatham, C. H., Depue, B. E., Banich, M. T., & O’Reilly, R. C. (2011). “A unified framework for inhibitory control.” Trends in Cognitive Sciences 15:10, 453–459.


Nicholls, A., Holt, N., & Polman, R. (2006). “Stressors, coping, and coping effectiveness among professional rugby union players.” Sport Psychologist 20:3, 314–329.

Oudejans, R. R., Kuijpers, W., Kooijman, C. C., & Bakker, F. C. (2011). “Thoughts and attention of athletes under pressure: skill-focus or performance worries?.” Anxiety, Stress, \& Coping 24:1, 59–73.

Pacherie, E. (2008). “The phenomenology of action: A conceptual framework.” Cognition 107:1, 179–217.

––– (2011). “Nonconceptual representations for action and the limits of intentional control..” Social Psychology 42:1, 67.

Pandey, S. & Sarma, N. (2015). “Utilization behavior.” Annals of Indian Academy of Neurology 18:2, 235.

Papineau’s, D. (2014). “Choking and The Yips.” Phenomenology and the Cognitive Sciences 14:2, 295–308.

Railton, P. (2009). “Practical competence and fluent agency”. In Reasons for action (ed. Sobel, D. & Wall, S.). Cambridge University Press, 81–115.

Rietveld, E. (2008). “Situated normativity: The normative aspect of embodied cognition in unreflective action.” Mind 117:468, 973–1001.

Ripoll, H. & Fleurance, P. (1988). “What does keeping one’s eye on the ball mean?” Ergonomics 31:11, 1647–1654.

Savelsbergh, G. J., Williams, A. M., Kamp, J. V. D., & Ward, P. (2002). “Visual search, anticipation and expertise in soccer goalkeepers.” Journal of Sports Sciences 20:3, 279–287.

Shepherd, J. (2014). “The contours of control.” Philosophical Studies 170:3, 395–411.

––– (2015). “Conscious control over action.” Mind & Language 30:3, 320–344.

Slingerland, E. (2003). Effortless action: Wu-wei as conceptual metaphor and spiritual ideal in early China. Oxford University Press.

Stanley, J. & Krakauer, J. W. (2013). “Motor skill depends on knowledge of facts.” Frontiers in human neuroscience 7, 1–11.

Stanley, J. & Williamson, T. (2001). “Knowing how.” The Journal of Philosophy 98:8, 411–444.

Sutton, J., McIlwain, D., Christensen, W., & Geeves, A. (2011). “Applying intelligence to the reflexes: Embodied skills and habits between Dreyfus and Descartes.” Journal of the British Society for Phenomenology 42:1, 78-103.

Velleman, J. D. (2008). “The way of the wanton”. In Practical Identity and Narrative Agency (ed. Mackenzie, C. & Atkins, K.). Routledge.

Wu, W. (2011). “Confronting many-many problems: Attention and agentive control.” Noûs 45:1, 50–76.

––– (2015). “Experts and deviants: the story of agentive control.” Philosophy and Phenomenological Research (published online).



[1] Intellectualism, thus defined, is arguably shared by most causal theories of action, from Davidson’s (1963) classical view to many recent ones (e.g. Pacherie (2008; 20011); Wu (2011); Shepherd (2014; 2015)). This definition differs from the most common meaning of ‘intellectualism’ in the skill literature, where it often refers to the view that knowing-how is not different from knowing-that, but rather a special subset of the latter (e.g. Stanley & Williamson 2001; Noë 2005; Stanley & Krakauer 2013). Accordingly, ‘anti-intellectualism’ is often taken to be the opposite claim that knowing-how is a kind of knowledge different from knowing-that. I use the terms in a different (though not entirely unrelated) way to try to capture the main views that have been proposed in the recent debate around skilled action.

[2] Nakayama’s statements are taken from the documentary Chef’s Table, directed by D. Gelb.

[3] Rietveld (2008); Brownstein & Madva (2012); Fridland (2015).

[4] See Slingerland (2003).

[5] E.g. Annas (2011, 17–18); Montero (2015); Wu (2015).

[6] Montero (2010; 2015); Sutton et al. (2011); Papineau (2014); Fridland (2014); Wu (2015). Many of these accounts respond to Dreyfus’ influential anti-intellectualist account of skillful coping (e.g. Dreyfus (2002), Dreyfus & Kelly (2007)).

[7] Sutton (2007); Di Nucci (2013); and especially Brownstein (2014) discuss these anti-intellectualist arguments.

[8] Velleman (2008); Railton (2009); Annas (2011).

[9] See also Dreyfus (2012).

[10] E.g. expert golfers who have been describing their technique need twice as many attempts to sink a putt than those who had previously performed an unrelated task (Flegal & Anderson 2008); expert golfers were more accurate when their attention was diverted from their own performance than when they were instructed to focus on it, while novice golfers presented the opposite pattern (Beilock et al. 2004).

[11] Term coined by Beilock & Carr (2001).

[12] Brownstein (2014, 555–6) reports these and several other expert testimonies indicative of expertise-induced amnesia.

[13] Beilock & Carr (2001). I assume a dual-process approach to human cognition, according to which automatic or intuitive processes are independent from, whereas reflective processes require, working memory. Notice that this dual-process approach is significantly different from coarser dual-system approaches. (On the general outlook, and the difference between dual-system and dual-process accounts, see Evans (2008; 2010); Evans & Stanovich (2013). On the concept of working memory, see Baddeley (2007).)

[14] Land & McLeod (2000).

[15] Fridland (2014).

[16] E.g. Logan (1985); Dreyfus (2002); Brownstein (2014).

[17] Annas (2011) and Montero (2015) make distinctions along these lines. Papineau (2014) proposes an analogous difference between “practice” (where the performer may automatically go along with the motions) and “competitive performance” (where the performer must keep her reflective focus on the task to seek any possible advantage).

[18] Baumeister (1984); Beilock & Gray (2007).

[19] Beilock & Carr (2001); DeCaro et al. (2011); Montero (2015).

[20] Beilock & Carr (2001).

[21] They define the contrast between novices and experts differently in several experiments. In this case, novices were participants with no previous golf experience, and experts were “local high school and college students with 2 or more years of high-school varsity golf experience or a Professional Golfers’ Association (PGA) handicap less than 8”. (2002, 1214).

[22] Nicholls et al. (2006).

[23] Oudejans et al. (2011).

[24] Moreover, there is evidence that this (i.e. using predictive saccades instead of directly following the ball) occurs not only in cricket (Land & McLeod 2000; Croft et al. 2010). Baseball (Bahill & LaRitz 1984), table tennis (Ripoll & Fleurance 1988), and soccer (Savelsbergh et al. 2002) performers seem to do it too.

[25] Land & McLeod (2000, 1342). See also Bahill & LaRitz (1984); Ripoll & Fleurance (1988).

[26] Participants were “Mark, a professional cricketer who has opened the batting for Warwickshire, Charlie, a successful amateur who plays Minor Counties cricket for Oxfordshire, and Richard, an enthusiastic but incompetent amateur who plays low-level club cricket” (2000, 1345).

[27] This is congenial to views recently developed by Wu (2011; 2015); Fridland (2014; 2015); and Shepherd (2014; 2015).

[28] Munakata et al. (2011) provide a conceptual framework to understand how the inhibition of irrelevant features and the selection of relevant features are two aspects of the same inhibitory process. Inhibitory control is a downstream effect of representing one’s current goal (a task associated with prefrontal-cortex areas), and it comes in two main varieties: the direct inhibition of neural activity in subcortical areas that would interfere with the task, and the direct excitation of goal-related neural activity in some areas which then indirectly inhibits potentially distracting neural activity by outcompeting it. This accounts for how the executive, intention-directed control of attention is crucial in both inhibiting irrelevant features and selecting relevant features, therefore making coherent, unified intentional action possible.

27 thoughts on “Do we reflect while performing skillful actions? Automaticity, control, and the perils of distraction”

  1. The paper offers an interesting contribution understanding the role that reflection might play in skillful actions, and I am sympathetic to the overall claim that is made in the paper. However, I think that the specific conclusion needs further clarification, especially on what role automaticity might play, what reflection precisely involves, and whether basic actions is the best concept to use here.

    The contrast between intellectualism and anti-intellectualism

    The difference between the intellectualist and anti-intellectualist position and what it precisely amounts to could be explained more clearly I think. Intellectualism is explained as the position ‘according to which action production requires high-order mental processes involving some state like intentions, beliefs, or desires’ (p. 2) and suggest that according to intellectualists skillful actions adhere to explicit rules (p. 6). This is contrasted with the anti-intellectualist view according to which the more skilled you are, the less you need to think about what to do. Also, this is related to whether there is a need to reflect in skillful actions, and whether automatic control might be sufficient for skillful action. I think that it is important to put the precise contrast between intellectualists and anti-intellectualists very clearly, and I am not sure whether this is sufficiently clear in this paper. First, I do not think that automatic control means that there are no mental states involved in performing the action, they do not necessarily exclude each other. Second, I do not think that reflection is a necessary condition for mental states to be involved in actions. At least, you have to explain more what you mean by automaticity and reflection to defend the distinctions you are making and to clearly delineate between the intellectualist and the anti-intellectualist position. Let me explain this in more detail.

    Automatic control and rational high-order processes

    Towards the end of the paper you conclude that ‘[f]rom this consideration of psychological evidence emerges the view that top-down, intention-directed, reflective mental processes do play a necessary role in the production of skilled action, namely, the executive modulation of attention’ (p. 20). I acknowledge that this focus of attention is important and that might follow from the psychological research, but I do not see how this focus of attention to the right aspects of the action is not itself something that to an important extent occurs automatically in skillful action by experts, especially since your own description of automaticity on p. 7 allows for this. For example, it seems to me that in general working memory is not involved in the defender’s attention being directed on the ball instead of the player while trying to prevent a goal being scored. It seems that focusing your attention on the right aspects of the action and keeping your attention on the action at the right level of description can be something that occurs automatically as well, without recruiting working memory. In the remainder of the paper the focus is on sensorimotor cases, which might obscure the fact that also cognitive subroutines can be automated through practice. It might very well be the case that, in general, experts do not have to reflect, or deliberate, or allocate working memory, or decide where to focus their attention on.

    Although this does not exclude your claim that intentions necessarily play a role even though attention is drawn automatically (I mainly want to highlight that using the term ‘automatic’ is problematic in this discussion), you do seem to suggest that the focus of this attention is something that occurs reflectively and not automatically, for example on p. 12 where it is stated that automatic control of action components ‘must be subordinated to higher-order reflective processes of attention guidance’ (p. 12, emphasis added). This seems to suggest that there is an active contribution on the part of the agent in guiding the attention in the right direction. Furthermore, to argue for this claim in more detail you show that experts report that they steer their attention actively, but these are cases in which the experts are under pressure. This does not show that in situations in which experts are not pressured (and not as susceptible to choking or the yips), they have to reflectively guide their attention as well. Therefore, it does not show that in general experts describe skilled performance as effortful (p. 16), and whether this is also the case in situations in which there is no pressure. This means that directing of attention might occur automatically in many situations in which experts act. Of course, you could still argue that an intention is needed to sustain the coherence between automatic guidance of attention and action control, but then it might be better to not take automaticity to be part of the anti-intellectualist position on skillful action and mainly take it to about whether skillful actions are agential (rational, minded) or not.

    Reflection and agential processes

    This brings me to my second point, which is the question what the precise relationship between reflection and agential processes is. From the paper it is not precisely clear what you mean with the term ‘reflection’, and I also could not grasp from the examples what is precisely meant by it in the paper and what the subjects are precisely reflecting on (e.g. are the subjects in the experiments reflecting or aware of the specific movements they make or of how they produce them, which mental processes are involved?). In related literature, the suggestion is that reflection is the involvement of conscious self-awareness or self-reflective thought (Brownstein 2014, p. 546). And from Gottlieb’s (2011, p. 347) discussion of Dreyfus’ view on reflection it is clear that he takes reflection to be about explicit judgments and ‘analysis and comparison of alternatives’. In the example of the finger movements (p. 7) this is also suggested: reflection is about having to explicitly relate letter strings to finger movements, in which we have to think about what we do. However, I think that intentions playing an explanatory role in action and whether the agent is reflecting on this mental state or analyzing and comparing courses of action can come apart. For an intention to play a unifying and cohering role and for an agent to be in this intentional state, she does not necessarily have to reflect or think about the intention. Think of examples in which having a certain intention does not need to be the result of any thinking process, but is the more or less direct result of environmental cues: the traffic light turning red, or the steps involved in cooking your favorite meal. This is precisely what the dilemma between intellectualists and anti-intellectualists is about, since anti-intellectualists might be deny that these routines involve intentions at all. My point here is mainly that invoking reflection makes the intellectualist view committing to more than it has to be. To argue against the anti-intellectualist one does not need reflection per se, because there might be a weaker sense in which intentions are still involved even without comparing alternatives or reflecting on the intention itself. Perhaps this is what you mean by reflection and then we might agree, but if that is the case this definition should be made more explicit and distinguished from cases in which we do not need to think or reflect about what to do but in which intentions still (seem to) play an explanatory role.

    Related to this, I think it is important to make a distinction between the actual involvement of working memory and the ability to reflect and report what you are doing and why when asked. The ability to report and to answer Anscombian questions might involve reflection, but that does not mean that the agent has to be reflecting at the moment of acting in order to still be able to answer these questions. Thus, even if the agent is not reflecting during the action, it might still be possible for her to answer Anscombian questions when asked. Gottlieb (2011, fn. 30) puts forth that this is precisely the difference between McDowell and Dreyfus: ‘The issue with Dreyfus is that conceptually informed actions must involve reflection in the moment or occurrently, whereas for McDowell, the concept involved must be available for reflection, reason, or in responding to reasons’. That means that McDowell, the intellectualist, allows for actual reflection during the action and the ability to reflect (and answer Anscombian questions) can come apart. This suggests that reflection in action is not necessary for an Anscombian intellectualist account (and this actually supports your view that these examples do not support anti-intellectualism); the agent does not have to hold the information to answer these questions in working memory at the moment of acting to be able to answer them.

    Of course, expertise-induced amnesia might still be a problem since that is about this ability to access the information. If convincing evidence for that can be found, that might still count as an argument against an Anscombian kind of intellectualism. However, it is also argued in the paper that the evidence for this is not convincing. This does make me wonder why you do not accept an Anscombian kind of intellectualism, because your discussion of the literature seems to allow for that as well. Given that, the question also is what is precisely involved in answering Anscombian questions. What should agents precisely be able to answer to and what is the level of specificity? Do they have to be able to report the rule they are following (e.g. p. 6) and to give descriptions of all the steps involved in putting (p. 14)? I am not certain whether that is what Anscombe had in mind; is that what counts as practical knowledge in action?

    Affordances and aims

    I also want to say something about your discussion of affordances and the soccer example. First of all, you criticized some of the experiments earlier for not studying actual experts, but then you do take yourself, not an expert, as an example of how attention is involved in skillful action. This undermines the claim you are making. Furthermore, I think the example is not that strong in some other respects. I do not think that in most cases of skillful action we have to reflectively inhibit other responses that are not suitable but instead that, because of the context of playing soccer, the situation simply does not afford these responses that do not fit. The context of playing soccer limits the affordances that occur, since some possibilities for action are simply not relevant in this context and therefore do not count as affordances. In relation to what I pointed out earlier, this also calls for clarity on what the position of the anti-intellectualist precisely is, because it seems to me that they (or at least some of them) do not deny aims playing a role, only not in the form of intentions. In relation, with affordances anti-intellectualists precisely try to explain how we can distinguish between relevant and irrelevant aspects of the situation without appeal to intentions, for example by the context or aims we have. The question of course is whether they are able to do so, but I think it is important to precisely argue against this claim. That is, why do we need an intention on the part of the agent for affordances to not be equally attractive? Why doesn’t the situation of playing soccer, in the case of the actual expert, suffice to explain this? This needs to be argued for in more detail. Furthermore, the distinction I put forth earlier, between mental states playing an explanatory role and reflection, is also of relevance here. The suggestion that in order to focus on the soccer-based affordances implies ‘reflective, inhibitory control’ might be strong. Focusing on the goal seems sufficient for our attention to be drawn to the relevant aspects of the situation and to not do something that does not fit this aim or goal. This makes the commitments of the intellectualist more stringent than they have to be.

    Attention to basic actions?

    Finally, the term “basic action” might be confusing here since its use is quite different from the way it is generally understood in the literature on action theory. There, basic actions are those actions that are done without doing anything else. As Lavin (2013, p. 275) explains: ‘a basic action is not the end of any other action; nothing else is done in order to do it; it is not an answer to “Why?” when asked about any other action’. In other words, it is those things that we simply do, for example moving your finger in order to flip the switch. It does not make sense to give the answer “to move my finger” to the question “Why?” In relating this to the goals of this paper, basic actions then precisely are not the aspects of the action that the agent should focus on. Therefore I think that the use of this term needs to be clarified and elaborated on more.

    Brownstein, M. (2014). Rationalizing flow: agency in skilled unreflective action. Philosophical
    Studies, 168, 545-568.
    Gottlieb, G. (2011). Unreflective action and the argument from speed. Pacific Philosophical
    Quarterly, 92, 338-362.
    Lavin, D. (2013). Must there be basic action? Noûs, 47(2), 273-301.

    1. Hi Lieke,

      Thank you for your detailed comments! I’ve grouped some of your concerns (particularly those about the concepts of intellectualism and reflection, and the possibility of a reflection-less skilled action) along with those of others in my reply below, but I wanted to address some of your more specific arguments here.

      About Anscombean intellectualism: I agree that Anscombeanism doesn’t require synchronic focus on the ‘why’s and ‘how’s of performance. But I would still resist Anscombeanism (as portrayed above) because it requires that action be performed on the basis of propositional knowledge. I’m not sure that the knowledge involved in skilled action has to be of a propositional nature.

      Concerning your argument that my use of myself in the soccer example invalidates the argument for attention-control intellectualism: I disagree. The idea would be: if even I, with my limited skills and in a relatively-low-pressure environment, cannot carry out a skilled action without the use of executive inhibitory control, this applies a fortiori an expert who has to perform under much, much higher levels of pressure and with many more sources of distraction.

      Finally, you suggest that practical situations have a stronger guiding power than I admit: “because of the context of playing soccer, the situation simply does not afford these responses that do not fit.” However, I can think of a few counterexamples, which suggest that, given sufficient pressure, even in a pretty unambiguous context, distracting forces can lead agents to behave in task-irrelevant ways. Think about Zidane getting kicked out of the World Cup final, in his last professional game, because he headbutted Materazzi after he said some insulting things to him… I can’t picture a more soccer-inducive context, or a more highly skilled performer, and still, he got distracted, stopped reflecting on his intention, got a red card, and lost the World Cup.

      This is only to suggest that, given examples like that one, maybe contexts have less of an action-guiding force. My general comments below argue for the more extensive need for reflection by reference to the phenomena of slips and utilization behavior.

      I hope the other key concerns are addressed below.

      Thanks again!

  2. Juan Pablo Bermudez ably defends a version of intellectualism about skilled action. His main claim is that the sort of “automatic control” anti-intellectualists describe is insufficient for explaining core features of skill in sports, the arts, cooking, and more. Intellectualism wins the day because to do these kinds of things well, one must exert reflective control over one’s actions. Bermudez argues for Attentional-control intellectualism. What makes skilled action intellectual, on this view, is not that the agent is acting on the basis of “knowledge-that” or propositional knowledge, nor is it that the agent understands what she is doing and why. Rather, what makes skilled action intellectual is the agent’s capacity to keep her attention focused on performance. Attentional-control intellectualism is the view that skilled action requires “the capacity to structure attention in relation to an intention.”

    My commentary will focus on various senses in which one might be said to reflect while doing things skillfully.

    Of the expert chef Niki Nakayama, Bermudez considers the claim that “her mind is not involved” when she is cooking. Presumably this is intended as a metaphor. Certainly Nakayama’s mind is active while she is cooking. If the brain is the mind, then Nakayama’s neural activity is enough to see this. Moreover, Nakayama is making decisions and having conscious experiences. These things require an active mind. So in what sense is Nakayama’s mind involved or not involved in her cooking?

    Nakayama herself says something which does not come up in Bermudez’s discussion. She writes, “Cooking is the one thing that I feel I can completely trust what I’m doing” (my emphasis). I take Nakayama to be speaking about the trust she has in her feelings, which is what she says guides her decisions when her mind is “shut off.” This is reminiscent of David Foster Wallace’s view of what great athletes do when they “shut off the Iago-like voice of the self.” Great athletes, Wallace suggests, don’t experience doubt or uncertainty while performing. This suggests (roughly):

    Trust: you are reflecting in skilled action when you fail to trust your feelings, habits, or spontaneous inclinations

    I will briefly come back to Trust later. As I mentioned above, Bermudez also leaves aside stronger forms of intellectualism, which suggest views like:

    Anscombeanism: you are reflecting in skilled action if you know what you are doing, why you are doing it, and you are acting on the basis of this knowledge

    StanleyandWilliamsonism: if you are acting skillfully, then you are reflecting, because knowing how to φ skillfully is knowing that some way w—the way you actually φ—is a way for you to φ

    And he also leaves aside a weaker form of intellectualism:

    Causal intention: you are reflecting in skilled action if your intentions are the causal antecedents of your motor behavior

    Instead, building on Wu (2011), Bermudez argues that intentions, which he takes to be no more than representations of goals, are “structural causes synchronic with action.” Skilled agents don’t just form an intention (e.g., to play a dropshot) and then run on auto-pilot (as Causal intention suggests). Rather, their intentions are active in action. Their intentions “modulate the activation levels of automatic association patterns.” Intending to play a dropshot primes “the relevant sensory, affective, and motor associations” while “the activation value of irrelevant associations is decreased.” Bermudez doesn’t offer details about how intentions do this, but the idea seems to be:

    Structural cause synchronic intention: you are reflecting in skilled action if your intentions are modulating your automatic reaction to stimuli while you are performing

    I get a little confused at this point in the essay, though. I have trouble putting Structural cause synchronic intention together with Bermudez’s core claim:

    Attentional-control intellectualism: you are reflecting in skilled action when you are focused on your intentions.

    There are two reasons I have trouble putting these together. The first is that intentions, as Bermudez defines them, seem sufficient for dealing with the problem of understanding how the skilled agent responds to relevant cues and ignores irrelevant cues (e.g., plays a dropshot but doesn’t yell at an obnoxious fan). Having the intention to play a dropshot seems sufficient for this since the intention is defined as modulating the agent’s response in an ongoing way, activating the right sets of responses and deactivating the wrong sets. So when Bermudez introduces the idea that attention is for keeping you focused on your intentions, I’m not sure what work attention is really doing. Bermudez writes, “by modulating the activation levels of automatic association patterns, the agent’s focus on her intention structures attention so that the relevant sensory features become salient, and hooked up with the relevant affective and motor routines” (my emphasis). Why does the agent have to focus on her intention in order to have these results, rather than simply have the intention? Presumably she intends to play a dropshot and doesn’t intend to yell at obnoxious fans. The definition of intending (in the synchronic sense) seems to explain how the relevant sensory features become salient and get hooked up with the relevant affective and motor routines.

    My second reason has to do with the causal sequence between attention and intention. Bermudez argues that attention is for keeping us focused on our intentions. But if intentions “structure the agent’s web of possible mental associations” that govern her behavior, don’t they structure her focus and attention already? Why don’t intentions govern attention, in other words?

    In a broader vein, I’d like to hear more about how, or to what extent, or in what way, Attentional-control intellectualism illuminates the role of the intellect in skilled action. I can see how views like Trust, Anscombeanism, and StanleyandWilliamsonism cleave in an intuitive way between the reflective and the unreflective. Whether true or not, these views operate with interpretations of “reflection” that seem genuinely intellectual. I’m less sure why Bermudez wants to count attentional control as intellectual or as sign of being reflective. Attention itself can be automatic, deployed without volition or awareness (Fridland 2014). Is it still reflective in this case? Or is it that some forms of attention are reflective and others aren’t? If so, what marks the distinction? And then, shouldn’t we be focusing on that?

    Here’s a manifestation of this broad issue. In responding to the anti-intellectualist argument from choking, Bermudez builds upon Papineau’s (2014) distinction between “choking” and the “yips.” Choking refers to performance failure due to “diverting reflective focus away from an action” (e.g., onto the crowd chanting “airball!”). The yips refers to performance failure due to “reflecting on automated components of an action” (e.g., thinking about how far your backswing should go when golfing). Bermudez argues that “reflecting on action components leads to loss of fine-grained automatic control, and ultimately to Yipping, while losing reflective focus on the basic action leads to distraction, and ultimately to Choking.” This is a helpful distinction, and it seems right that these are two distinct kinds of performance failure. (Although it might be good to hear why Bermudez rejects the view — in, for example, Poldrack et al. (2015) — that as skill increases, attention to what one is doing decreases.) But is choking a result of having the wrong reflective focus or of having the wrong focus simpliciter? What work is “reflective” doing here? Bermudez says that the advice to a choker might be to “keep her eyes on the ball and her head in the game.” Is this tantamount to thinking about the game in a reflective way? Keeping one’s head in the game seems like a good thing to do, and I can see how it demands attentional control, but I’m not sure I see why it should count as a reflective or intellectual activity. If choking results from having the wrong focus simpliciter, then a solution for choking might be to change one’s focus, like one changes one’s grip on the racquet. What’s intellectual about that? Similarly, Bermudez writes, “the more pressure there is on the agent to perform, the more reflective control seems to be needed.” But here’s an alternative: you might cultivate an attitude of not caring about the pressure. Then you could perform well without needing to try hard to control your attention, right?

    Maybe one way to approach the question of what counts as reflection is by considering what we are trying to do when we inquire about reflection in skilled action. Are we:
    • trying to model the taxonomy of the mind, and to understand whether knowing how is different from knowing that?
    • trying to understand whether experts are different in degree or kind from competent performers and novices?
    • trying to figure out how to teach and attain expertise?
    • trying to understand if skill in physical activities is different in kind from skill in mental activities?
    • aiming to use skilled action as a counterexample to “reflectivist” theories of action and agency?
    These questions are all related, but the relevant concept of “reflection” might change depending on which question we are trying to answer.

    1. Hi Michael,

      Thanks so much for your comments! (As you can see, I’ve used the distinctions between StanleyandWilliamsonism, Anscombeanism, etc., to set up the debate in the video.)

      I’ve gathered what I took to be your main points and offered answers to them in the general reply that I’ve posted below. But here I wanted to consider some of the more specific points you’ve raised.

      You say that, according to my paper, “attention is for keeping us focused on our intentions”. I would rather say that reflection is for keeping us focused on our intentions. Or, in other words, that reflection is for keeping our intention active and sustaining its attention-structuring power throughout performance. In the comment below I say that reflection is any higher-order mental process that recruits working memory, i.e. executive functions. It’s one of the key executive functions (often labelled ‘inhibitory control’) to keep a representation of one’s goal active. This has the double effect of priming the available goal-related information and inhibiting the available goal-unrelated information. (You’re right that I don’t discuss details of how this occurs; for more detail I rely on Munakata et al. (2011).)

      So I grant that it should be the other way around: that “intentions [should] govern attention”. Still, we must be able to explain why in some cases intentions are govern attention more effectively than in others (e.g. in cases of slips or Choking). And this, I think, is explained by reference to reflection—or the lack thereof.

      On your suggestion that, instead of exerting more executive control, the skilled agent “might cultivate an attitude of not caring about the pressure”: certainly a relaxed attitude would help in performing skilled actions. It may in fact be necessary for performing at a very high level that you learn not to care that much about all that pressure. But I don’t see the development of this attitude as an alternative to exerting inhibitory control over your attention. This is because even in cases of low pressure reflective control over your attention is necessary for skilled action performance. I argue for the latter claim in points 2 and 3 of my general reply below.

      Your comments on Trust are very suggestive, thanks for pointing out that aspect of Nakayama’s claims. I’d say that reflecting, on the one hand, and trusting your feelings or spontaneous inclinations, on the other, needn’t be mutually exclusive. In fact, perhaps both are necessary for skilled and expert action. Otherwise I don’t see how those feelings and inclinations could be trained and refined further…

      Thank you again for your thought-provoking comments.

      1. Thanks! This helps me understand your view. Now it seems to me that a key question will be whether it is true, or under what conditions it is true, that reflection, as you define it, does indeed keep us focused on our intentions. My sense is that this will come down to the definition of reflection, and of automaticity, as some of the other commentators have suggested. One worry I might have is defining reflection so broadly as to make any action reflective. This is why I think it matters what question we are trying to answer (in the way I said in my commentary). You say in your reply to everyone (below) that reflection in your broad sense should count as intellectual. I guess I think that depends on what we’re talking about. Maybe it’s intellectual in the context of the knowledge-how debate; maybe it’s not intellectual in the context of distinguishing experts from competent performers or in the context of distinguished action types (e.g., deliberative action and spontaneous action).

        On trust: I think there may be more of a tension between trusting your spontaneous inclinations (as in the Nakayama example) and reflecting on your intentions. I’m tempted to think that, in some situations at least, reflection undermines this kind of trust. If I am playing tennis, and am deliberating about whether to try to serve a big bomb up the middle or to play it safe with a kick serve, I am precisely not trusting my spontaneous inclinations. But maybe this isn’t the kind of situation you are thinking about. Or maybe this is a specific kind of reflection, which is different from the kind you are talking about. This connects with the last point above.

        1. I agree with you that

          “a key question will be whether it is true, or under what conditions it is true, that reflection, as [I] define it, does indeed keep us focused on our intentions”.

          That said, this sounds to me like an empirical question, rather than a matter of definition. The answer would certainly change if we vary the definitions of ‘automaticity’ and ‘reflection’. But we may be able to advance empirically toward an answer if we keep the definitions constant (by reference to executive functions), and ask: when, and how, does impairing different executive functions harm our ability to coordinate behavior toward attaining a certain goal?

          The issue about labelling my minimal notion of reflection an intellectual process does sound much more like a verbal dispute, or relative to the question we want to answer (and this leads back to Josh Shepherd’s worry that it may just be a battle of words). I’ll have to think more about that.

          On your final point: I agree that the kind of reflection in your example would be incompatible with trusting your spontaneous intuitions. But if the proper description of what you’re doing is ‘playing tennis’, it seems to me a case of reflecting on action components rather than on the action itself. And, yes, I’d say it is a specific type of reflection that goes beyond sustaining the activation of your goal’s representation.

          1. Yup, totally agree that it is an empirical question whether, or under what conditions, reflection keeps us focused on our intentions. I’m less certain than you about how to interpret the extant studies (e.g., the “watch the ball” case), but more and better studies are needed in any case.

            One other point: I’m not sure the proper description in my example is “playing tennis.” There are many possible descriptions. Heading in the direction of greater and greater generality, I am playing tennis, playing a game, keeping fit, moving my body, etc. And heading in the diretion of greater and greater specificity, I am playing the second set, playing my service game, serving, trying to hit an ace, trying to hit an ace up the middle, trying to serve with a deep knee bend, trying to flick my wrist in the “this” way, etc. It’s not clear to me how we pick the right action-description, or how we decide where to cut between action-descriptions and descriptions of components of action. The case of interest to me is the one in which the expert reflects upon whatever it is that only she does (i.e., the thing that distinguishes her from the ordinary performer). This wouldn’t be “playing tennis,” I assume.

            1. Michael:

              Just really a high-five comment. You say:

              One other point: I’m not sure the proper description in my example is “playing tennis.” There are many possible descriptions. Heading in the direction of greater and greater generality, I am playing tennis, playing a game, keeping fit, moving my body, etc. And heading in the diretion of greater and greater specificity, I am playing the second set, playing my service game, serving, trying to hit an ace, trying to hit an ace up the middle, trying to serve with a deep knee bend, trying to flick my wrist in the “this” way, etc. It’s not clear to me how we pick the right action-description, or how we decide where to cut between action-descriptions and descriptions of components of action.

              ***The case of interest to me is the one in which the expert reflects upon whatever it is that only she does (i.e., the thing that distinguishes her from the ordinary performer). This wouldn’t be “playing tennis,” I assume.


              I think that’s the most important case to get clear on. It seems to me that a lot of talk about intentions in skilled action doesn’t really address this ***last point.

              1. I agree with both of you: that’s the crucial case. Also, because the same action can be described in potentially infinite ways, I find resorts to action descriptions frustrating…

                One question about Michael’s ***last point: could it be that what distinguishes the expert from the ordinary performer is not whatever it is that she reflects upon? My intuition is: it may be that both performers reflect on the same thing (e.g. ‘Keep your eyes on the ball’), but that the expert’s fine-tuned automatic routines allow her to implement this in a different way (like, e.g. performing two predictive saccades instead of just one). The difference thus may be, not in the content of her reflection, but in the way that her chunked behavior has been parsed through practice…

  3. Thanks to Juan Pablo Bermúdez-Rey for a very interesting paper. There is a lot one wants to say, but I will confine my commentary to two issues. The first concerns Bermúdez-Rey’s set-up of the debate. The second concerns his positive view.

    Bermúdez-Rey sets up the debate as one between Intellectualists and Anti-Intellectualists. For philosophers discussing skilled action, this is a pretty familiar way of setting things up, although depending on the philosopher, small differences in the way each position is described could be chronicled. Here is how Bermúdez-Rey articulates each position. Intellectualists hold that ‘action production requires high-order mental processes involving some states like intentions, beliefs, or desires’ (p. 2). Anti-intellectualists hold that ‘the more skilled you are, the less you need to think about what to do or how to do it; the practical situation simply solicits the proper response from you’ (pp. 2-3).

    As stated, these positions might be in tension. But they are not in direct conflict. I doubt that every Anti-intellectualist will want to deny that beliefs, desires, and/or intentions are absent in skilled action. Further, one can hold that action requires states like intentions while maintaining that the more skilled one is, the less one needs to think about one’s action. Much seems to hinge on how we are meant to understand notions like ‘thought about action,’ ‘high-order mental processes,’ and ‘solicitation of proper response.’

    Ought we to conceive of discussion regarding skilled action as best conducted in terms of a debate between Intellectualists and Anti-intellectualists? My own view is that though something like this debate exists in the literature, there is more room here for merely verbal dispute than substantive debate. Acknowledging this will have ramifications for how we think about the nature of skill – which, as Bermúdez-Rey notes, is ‘part of what is at stake here’ (p. 7). Bermúdez-Rey notes that Intellectualists tend to understand skill in qualitative terms: you either have it or you don’t. Anti-intellectualists tend to see skill as a graded phenomenon. The connection between a qualitative view of skill and Intellectualism makes sense if you take the Intellectuals to require the presence of high-order mental processes in every instance of skilled action control. But Intellectualists ought not require this much. It is more plausible to maintain that high-order mental processes are important for some action-types (e.g., ones the execution of which is fairly complicated), especially in some circumstances (e.g., circumstances novel or complicated enough to preclude the usefulness of full automatization of action components). If this is what you think – and this is what I think – then you can be an Intellectualist of a sort without denying the insights of Anti-intellectualism, and while maintaining the plausible view that skill comes in degrees.

    The second issue I wish to raise concerns Bermudez-Rey’s positive account. He argues that reflection has an important role in skilled action control, and that the element of reflection that is important is attention. He notes that this view is congenial to views developed by Wayne Wu, Ellen Fridland, and myself. This is true, but in part for this reason focusing on differences may be illuminating. There are three short points I wish to make.

    First, Bermúdez-Rey’s view is that ‘top-down, intention-directed, reflective mental processes do play a necessary role in the production of skilled action’ (p. 20). This seems too strong to me. At the very least, I am not sure we have eveidence in favor of a necessity claim here. Admittedly, talk of necessity and its denial gets us into questions about the nature of skilled action, and these questions are hand in hand with others about the nature of intentional action and its causal structure. So I will not do more here than raise the possibility that cases of skilled action without reflective mental processes exist. If that is right, Bermúdez-Rey should consider softening his view on the relation between reflection and skilled action.
    Second, Bermúdez-Rey has it that the way reflective processes assist skilled action is via reflection on one’s intention – one must ‘attend to the representation of [one’s] current goal’ (p. 21). This seems overly restrictive. There are many things to which one might attend while acting. One’s intention might be one of them. (Parenthetically, however, it bears mention that some will argue that this gets things the wrong way round. According to some – including Bermúdez-Rey it seems – intention structures attention. One natural reading of this has it that attention is then directed at features of the world, not back at the thing structuring it.) But why exclude a role for attention to anticipated events, to bits of the perceptual array, to the level of effort one is currently deploying (especially if one senses a need to try harder), etc.?

    Third, Bermúdez-Rey’s ‘attentional-control intellectualism’ is, as noted already, set against an Anti-intellectualist view that emphasizes automaticity. But I wonder whether, for at least some action-types, the kind of attentional processes in play could not themselves become automatized, or at least partially automatized. If so, returning to my first comment, we might not need to oppose Bermúdez-Rey’s Intellectualism with Anti-intellectualism. Rather, we might combine the points of emphasis important to both views. That is, we might have a view that emphasizes the Intellectualist point that higher-order mental processes are important for action control, while granting the Anti-intellectualist insight that in many cases the structure of action control has more to do with learning and automatization than with free-form conscious reflection.

    1. Just to add some thoughts to Josh’s points. I enjoyed the video presentation of Bermúdez-Rey though I did not read the paper so sorry if I haven’t taken in the nuances and qualifications. What work is “reflective” doing here. Since my account came up as a possible ally, I would add that on my view, any intentional action involves the influence of intention in setting attention. In perceptually guided action, that means that intentions set perceptual attention, always. If that interaction is enough to count as a form of reflective influence, then I would say that the necessity claim is correct…but it might not be the sort of intellectual effect that the intellectualists countenance (in a different debate, I argue that perceptually-guided action involves pervasive cognitive penetration, and it would be interesting to see how that debate differs from the current one about “intellectualism”).

      Second: I think you need an account of automaticity and control, to clearly talk about these points. “Automaticity” should be a technical notion in action theory but we’re still using it in a non-technical way that relies too much on intuition and common usage. Without this, I think any discussion will be hampered. Yet the literature on this is pretty thin despite the pervasive use of the automaticity/control distinction or similar ideas.

      Since I don’t really know of other detailed definitions out there (Ellen, Josh, do you folks have one?), on my view, you specify automaticity relative to some action-property, and an action is automatic relative to that property if the action is not generated by an intention to act in that way. Otherwise, if there is such an intention, then the action is controlled relative to that property. THere is no such thing as a controlled or automatic process simpliciter unless one quantifies across action-properties. The mistake in the empirical/philosophical literature is to talk about automaticity and control without care as to what it might come to (Ellen has a nice paper detailing empirical work on this topic and I’ve given a briefer overview of it in a paper on Mental Action).

      1. Thanks again to Josh, and to Wayne, for very helpful comments.

        Josh and others have helped me realize that it’s better to do away with the talk of ‘attending to one’s intention’. You’re right that attention should be directed at things other than the representation of one’s goal. That said, what I still want to capture is (1) that, in order for skilled action to take place, the representation of one’s goal should modulate the salience levels of the possible objects of attention. And (2) that this modulation requires the goal’s representation to stay active during performance—because otherwise distraction may lead to performance breakdowns like slips or Choking. So now I think a better way to express this is to claim that it is necessary to reflect on one’s goal, in the sense that executive functions must sustain the activation of the goal’s representation, so that the representation may modulate salience levels (through inhibitory control).

        This latter point can be linked up with Wayne’s claim that, on his view,

        “any intentional action involves the influence of intention in setting attention”.

        I would add to this that, for intention to play this influencing role, its influence has to be sustained throughout action performance. Hence the need to recruit executive functions in skilled action. (I try to spell this out further in the general reply below.)

        —Josh’s worry that this is just a verbal dispute is still worrying me. I’ll have to think more about this.

        —I share Wayne’s view that a precise characterization of ‘automatic’, ‘reflective’, and ‘controlled’ are crucial for this debate to be fruitful. Ellen’s (2014) and Wayne’s (2013) papers are certainly helpful in that regard, and Brownstein and Madva’s (2012) should also be mentioned. I’ve been working on defining these notions elsewhere (and can’t develop my view in sufficient detail here). My strategy differs from Wayne’s in that I try to provide a definition of automaticity that isn’t relative to an action-property description, and in that I reject the ‘simple connection’ between the presence of control and the absence of automaticity. My main strategy is to define ‘automaticity’ and ‘reflection’ by reference to executive functions: any mental process employing them counts as reflective, and any process that doesn’t counts as automatic. ‘Control’ cuts through this distinction, however, given the normative character of many automatic processes (Brownstein & Madva 2012), and evidence that automatic processes have reflection-independent error-correction mechanisms (e.g. Rabbitt 2002).

        My reply to all (below) tries to address other issues Josh and Wayne have raised.

        Thanks again!

        Brownstein, M. & Madva, A. (2012). “The normativity of automaticity.” Mind & Language 27:4, 410–434.

        Fridland, E. (2015). “Automatically minded.” Synthese first online. doi:10.1007/s11229-014-0617-9

        Rabbitt, P. (2002). “Consciousness is slower than you think.” The Quarterly Journal of Experimental Psychology: Section A 55:4, 1081–1092.

        Wu, W. (2013). “Mental Action and the Threat of Automaticity”. In Decomposing the Will (ed. Clark, A., Kiverstein, J., & Vierkant, T.). Oxford University Press, 244–61.

        1. Jaun Pablo, can say a little bit about what you think the connection between automaticity and control is?

          1. I’ll try…

            My main view is that there are two independent kinds of control: automatic control and top-down, executive control. I think we have evidence for this from studies like Rabbitt (2002), which I mentioned above. Here’s the shortest description I can manage right now:

            Rabbitt compared people’s abilities to correct, report, and recall errors in a simple task. Participants were instructed to look at a screen split into four squares; when a dot appeared in each of the squares they had to press a corresponding button. One group of participants was instructed to immediately correct the mistakes they made during the task; a second group was instructed to press a fifth button every time they recognized they had made a mistake; a third group was randomly interrupted and asked whether they remembered having made a mistake in the last three responses; and a fourth group was told to ignore all errors and just keep responding. The length during which the stimulus appeared (i.e. the Response Signal Interval, [RSI]) varied randomly throughout the task between 150ms, 500ms, 800ms, and one second.

            The first relevant finding was that for all participants the response that followed a mistake was slower than the previous ones. This slowing occurred even when they were unable to report or recall the errors. Further, participants in the first group were remarkably fast and accurate in correcting their errors: throughout all RSI’s, errors were on average faster than correct responses, and error corrections were on average even faster than errors.

            Additionally, participants displayed much more accuracy in error correction than in error report or error recall. Across all conditions, it took participants much longer to report an error than to correct it, and failed to report errors much more often than they failed to correct them. Given that error correction can occur notably quickly (as quickly as 40ms after the error is performed), Rabbitt concludes that error correction cannot depend on executive recognition and rectification processes. The error-correction mechanism must therefore be automatic. Crucially, these automatic error corrections always worked to increase response accuracy, and never turned a right response into a wrong one.

            —This, and several other sources of evidence, lead me to I think that automatic processes can be sensitive to error, and able to perform quick and efficient self-correction on their own, independently from executive functions. This supports Michael and Alex Madva’s view that automatic processes display a “[r]esponsiveness” that is “flexible, self-modifying, and […] a genuinely normative phenomenon” (2012, 428).

            In another paper I’m working on I build a case for the view that such automatic responsiveness is different and independent from top-down, executive control. (This is work in progress, but I’d love to share it when the draft is somewhat presentable.)

            Brownstein, M. & Madva, A. (2012). “The normativity of automaticity.” Mind & Language 27:4, 410–434.

            Rabbitt, P. (2002). “Consciousness is slower than you think.” The Quarterly Journal of Experimental Psychology: Section A 55:4, 1081–1092.

  4. Hi Juan Pablo,

    Very interesting paper:

    I’d like to second interest in hearing more about a few issues that were raised in comments above:

    (1) What do you mean by “reflective” and what work is it doing for you?
    (2) Why think that attention to an intention is necessary for the intention to structure an action (as opposed to holding that the intention structures attention (as both Michael and Josh have pointed out)?
    (3) Why think of your position as Intellectualist?


    1. Hi Ellen,

      Thank you for distilling some of the key questions raised by the other commenters. I’ve tried to answer all of these below.

  5. Hi Juan Pablo,

    Super interesting paper!

    I just wanted to sound a note of caution this in the debate about the label ‘Anscombean Intellectualism’. There’s a boring point here, which is this view seems un-Anscombean (at least to me!), and a more interesting point about how Anscombe’s notion of practical knowledge plays into skilled action.

    As I understand Anscombe, she claims that intentional action requires practical knowledge of what one is doing, and why one is doing it at the time of action. She explicitly distinguishes this kind of knowledge from observation, likening it to the ‘maker’s knowledge’ had by someone who is designing a building. Now it’s a super-difficult question how to understand the possibility of this kind of knowledge, but there is a swathe of contemporary action theory which takes the idea that intentional action requires this kind of practical knowledge seriously.

    By contrast, in this paper (and in Michael’s commentary), the label ‘Anscombean Intellectualism’ seems to be being used to pick out the idea thatl action involves the ability to answer the question ‘what are you doing’, via reflecting on what one is doing. For example, Juan says:

    ‘Brownstein argues that the reflective self-awareness necessary to answer Anscombean questions is precisely the kind of over-thinking that causes yipping.’

    And Michael characterises Anscombeanism in the following way:

    Anscombeanism: you are reflecting in skilled action if you know what you are doing, why you are doing it, and you are acting on the basis of this knowledge.

    There seems like there is a perfectly good view here (according to which what and why questions get answered by reflection on action), but the label is unfortunate, since Anscombe claims that the ability to answer what and why questions is grounded in non-reflective practical knowledge. That’s the boring point about the label. The more interesting point is to think about the role of non-observational practical knowledge in skilful action. If we can make sense of the idea of non-reflective practical knowledge, then we might ask whether skilful action involves this kind of practical knowledge, how practical knowledge interacts with self-reports (especially self-reports after the fact, note the careful time indexing above!), and what this tells us about the role of thought and knowledge in skilful action.

    What do you think? Am I missing something about how ‘Anscombeanism’ connects up with Anscombe’s view?

    Anyway thanks again for the paper, and really interesting commentaries!

    1. Hi Josh,

      On the boring point: I used the term in my “Rationalizing Flow” paper that Juan Pablo discusses in the context of Peter Railton’s claim that, “were we foolish enough to interrupt [experts in e.g., sports and the arts] in mid-stream, they could typically answer the Anscombean question, ‘What are you doing?,’ without further observation or inference.” I adopted Railton’s usage of “Anscombean” in order to argue against his claim. As I say in the paper, I had nothing at stake in this being an accurate characterization of Anscombe’s views. But yes, sounds like the label is a bit problematic.

      On the interesting point: I’d love to hear if Anscombe has interesting ways of answering the question you asked (i.e., if there is an account of non-reflective practical knowledge that helps to explain the relationship between skillful action and self-reports)!

      1. Hi Michael,

        That’s really helpful on the origin of the label. So I guess the important point is to distinguish:

        WEAK ANSCOMBEANISM: an agent can say what they are doing and why while they are doing it, without needing to find out

        STRONG ANSCOMBEANISM:weak Anscombeanism is true, plus this knowledge-what and knowledge-why is practical knowledge rather than observational knowledge

        Weak Anscombeanism isn’t per se problematic as a label (since it is a view that Anscombe subscribes to), but the combination of weak Anscombeanism with the idea that the knowledge-what and why is reflective (hence the denial of strong Anscombeanism) is not a very Anscombean view.

        On how to explain the cases of agents expressing false beliefs, Markos’ suggestions below connect really close with what I was going to say. I think that there’s basically a family of responses to be made here:

        1) Agent’s knowledge is possessed whilst acting, but there’s no reason to think it is possessed after action has finished: as Markos suggests, it might be transitory. This idea I think is bolstered if we reflect that so-caled skilled actions have quite a complex intentional structure which will presumably be rather difficult to keep track of. (this might be the right response for the cases of people who say that they have no idea what they are doing)
        2) As Markos suggests Agent’s knowledge is compatible with other faulty routes to forming beliefs about actions. That you know what you are doing, doesn’t mean that you can form faulty contrary beliefs to the contrary.
        3) Agent’s knowledge is only of the descriptions under which action is intentional, so don’t expect people to know other descriptions of what they are up to (One of these two responses might work for the cricket confabulation cases).

        1. Cool, thanks Josh. This is helpful.

          On the idea that knowledge is present while acting but that there’s no reason to think that it is possessed after the action has finished: this seems plausible to me. I like the way Markos put it: that maybe the confabulating expert knows what she is doing in one sort of way while failing to know what she is doing in another sort of way. Accepting this inclines me toward anti-intellectualism, since, as Markos says, we are positing two different ways that the agent knows things.

          1. Thanks for these remarks on Anscombeanism, guys.

            On the label: I used it merely following Michael, the same way he was following Railton! It’s good to have a more precise characterization of Anscombe’s actual view.

            About Markos’ and Josh’s suggestions on expert confabulation: Your refined characterization of Anscombeanism would indeed offer ways to explain the phenomenon. And the explanation would veer towards anti-intellectualism, like Michael suggests. But, as I argue in the paper, there really is no strong evidence that expert confabulation exists! At least when it comes to the ‘Keep your eyes on the ball’ research, experts don’t actually have false beliefs: they really do track the ball with their gaze, so their beliefs seem to be true.

            So there’s no need to apply Anscombeanism to the explanation of this phenomenon: the phenomenon doesn’t seem to exist. (At least, further evidence is needed to show it does.)

            Rather, the interesting thing is that, according to the evidence, experts’ knowledge whilst acting is possessed by them after the action has finished. This inclines me toward intellectualism, because at least a portion of the expert’s knowledge in-action is available to her for later self-report, which suggests that there is some continuity between that knowledge-in-action and the mechanisms that produce self-reports.

            Of course, not all of the expert’s knowledge in-action is available for self-report. Specifically, knowledge about her automatized action components (to use Papineau’s distinction) is probably unavailable post hoc (the expert batter may not know the details of how she directs her eyes toward the ball, with how many saccades, etc.). But, crucially, knowledge in-action about basic actions does seem to be available post hoc (she does know that she has to keep looking at the ball, as do trainers and practitioners).

            A similar intellectualist reply seems available for expertise-induced amnesia: expertise generates amnesia about action components, but doesn’t seem to generate amnesia about basic actions, because experts know what they have to do to perform at the top level.

            1. Hey Juan Pablo – I’m curious how you interpret the Mann and colleagues study (2007) in which they had skilled cricketers bat while wearing misprescribed contact lenses. Their ability to bat was only significantly affected once they were wearing lenses so distorting that their eyesight was on the border of legal blindness. Doesn’t this suggest that these batters didn’t need to watch the ball clearly in order to bat? (A note of caution on this study, though: it had a whoppingly large n of 11!)

  6. I want to thank the organizers for their incredible work in putting this event together (even despite a hurricane!), and the commenters for their generous and acute texts, which have certainly helped me reassess my views. To continue the discussion, I gather their remarks in four main questions:

    (1) What is reflection, and what is its role in my argument?
    (2) Why is focusing on an intention necessary for performing skilled actions?
    (3) Can’t there be simple, low-pressure skilled actions that don’t require reflection?
    (4) Why should my position be called ‘intellectualism’?

    I offer my replies to each of these below, trying to clarify things that should have been clearer, and to address the core issues raised.

    1. Reflection and its role in the argument

    In the paper (note 13) I briefly mention that my notions of ‘reflection’ and ‘automaticity’ are taken from recent developments in the dual-process framework, particularly from the work of J. St. B. Evans (2008; 2010), who holds mental processes to be classifiable as either intuitive or reflective. Reflective mental processes are those that recruit working memory, broadly understood (i.e. in as coextensive with the executive functions (Baddeley 2007; Diamond 2013)). Thus we have:

    Reflective process: Any mental process that requires the recruitment of executive functions.

    Automatic process: Any mental process that does not recruit executive functions.

    Traits of reflective processes that follow from this are that (1) they require some cognitive effort, since executive functions are costly mental operations; that (2) their role as higher-order mental processes is to guide and control lower-order, automatic processes toward the completion of a goal; and therefore that (3) they imply the sustained activation of the goal’s representation.

    In the paper and the video, I argue that skilled action is a reflective, intellectual exercise because its performance requires reflection. Given its necessary reflective component (more on the necessity claim below), skilled action has a key trait in common with other intellectual phenomena (like solving intellectual puzzles, resisting temptation, and planning), namely that they cannot be performed without the recruitment of executive functions.

    Thus, what reflection does in my argument is it allows me to pinpoint the intellectual component of skilled action, which in turn allows me to hold skilled action to be an intellectual process.

    2. Why must we focus on our intention?

    Why isn’t having an intention sufficient for synchronic agentive control? I think we have to distinguish between having an intention(which I understand very broadly as having the representation of a goal) and reflecting on that intention (i.e. the mental process of keeping the goal’s representation active so that it can structure downstream, lower-order processes). We can have an intention, and nevertheless fail to keep the intention active. This is what happens in the case of everyday slips (like e.g. when I leave the office intending to stop by the pharmacy on my way home, but end up driving home instead), which I briefly mention in the paper.

    Slips are key here because they are cases in which having an intention is not sufficient for structuring our attention and controlling our action. In slips the agent’s governing intention cannot effectively structure attention, and this leads to a breakdown of performance that is manifested in the agent’s acting contrary to the current governing intention (Amaya 2013). I argue in the paper that Choking (in Papineau’s sense) is analogous to slips, in that both are performance breakdowns that take place because we fail to reflect on our intention during performance.

    If this distinction between having an intention (having the representation of a goal, which may or may not be active at any time) and reflecting on an intention (sustaining the representation’s activation so that lower-order attentional and behavioral processes can be structured in accordance with it) holds, it becomes clearer why merely having an intention is not sufficient for performing skilled actions. Reflection is also necessary.

    Now, it may seem that reflection is necessary only in high-pressure situations, where the probability of distraction is high or the agent is attempting to do something notoriously complex (I take this to be Josh Shepherd’s proposal in his comment). But the problem of distraction seems to be more pervasive than this. Slips themselves illustrate that intentions can become ineffective in everyday situations. Another version of my paper included a more developed argument (which now I see I shouldn’t have removed) for the claim that executive functions must be recruited even in performing simple, habitual, everyday skilled actions, from tying our shoelaces to driving to work. I’d like to briefly sketch this argument as a provisional reply to the following question.

    3. Can’t there be low-pressure skilled actions that don’t require reflection?

    Consider the psychological phenomenon called ‘utilization behavior’. Utilization behavior patients are like people whose entire conduct is constituted by slips: they respond immediately to any of the situation’s present affordances, but seem to have lost any sensitivity to broader practical concerns (L’Hermitte 1983; Iaccarino et al. 2014). Thus, a patient may continuously switch a light on and off; if she sees a comb she may start combing her hair, regardless of who owns it; if she is in a bedroom she may undress and go to bed, even if she is just visiting; upon noticing a coffee cup she may start drinking from it, although it was the experimenter’s coffee; and so on.

    The cause of utilization behavior seems to be that the condition’s associated neural damage disrupts the system in charge of generating endogenously-driven behavior (Rietveld 2012). The patient can properly respond to many exogenous affordances, but she cannot modulate these responses on the basis of internally generated priorities or motivations. This lack of the ability to endogenously modulate conduct leads her to behave in erratic ways that hinder even simple, habitual skillful coping with her surroundings. Notice, however, that utilization behaviors are in a sense correct: they involve skilful manipulations of objects (e.g. patients do not drop the coffee cup, but sip from it appropriately) which display successful automatic control over behavioral routines. But rather than calling these routines ‘actions’, it seems more appropriate to call them mere reactions to the environment’s affordances; exogenously-driven automatic responses that lack the internal coherence proper to intentional actions. The behavior of these patients seems not to constitute full-fledged action because they are missing two things: (1) the capacity to generate endogenous intentions, and (2) the ability to sustain a single intention throughout a significant amount of time.

    So what distinguishes a normal agent from a utilization-behavior patient is the former’s ability to form intentions and to reflect on intentions (so that the intentions can successfully structure attention and conduct). Thus, through the cases of slips and utilization behavior, I try to give traction to the claim that reflection is necessary for skilled action. This argument, of course, requires much further development, but that’s a gist of how I would go about answering the question.

    4. Why call my position ‘intellectualist’?

    In the context of the skilled action debate, calling such an admittedly minimal view ‘intellectualism’ does seem odd. The label ‘intellectualism’, after all, is associated with the stronger positions of Stanley&Williamsonism and Anscombeanism, as Michael has portrayed them. However, I think (and I claim as much in the video) that we should conceive of intellectualism as including a much broader family of positions. Not all intellectualism is propositional intellectualism (i.e. it doesn’t have to hold that reflecting on propositions is necessary for skilled performance). There is also a substantive intellectualist position that holds that skilled action requires reflecting, although this reflection does not consist in the manipulation of propositions.

    The issue then turns on whether reflection, understood broadly as ‘the use of executive functions’, is a sufficiently cognitively demanding process to merit the label ‘intellectual’. At this point I tend to agree with J. Shepherd that the debate runs the risk of becoming a fight over words. But here’s a way in which we may draw a substantive line between intellectualism and anti-intellectualism. I would call ‘intellectualist’ any view that argues that some necessary aspect of skilled action is not fully automatable. In other words, anyone who holds that skilled action requires some mental process that is not susceptible to automatization would count as an intellectualist. Anyone who says that some skilled actions can be performed fully automatically would count as an anti-intellectualist.

    Now, given the standard dual-process definitions of ‘reflection’ and ‘automaticity’ I’ve outlined, it seems clear that reflective processes are not automatable. Executive functions are simply not the kind of thing that is susceptible to automation. As several commenters pointed out, the processes of directing attention to task-relevant features are susceptible to automation and refinement; but the process of sustaining the activation of the goal’s representation seems not to be susceptible to automation. It seems to necessarily require the recruitment of executive functions.

    Thus, my view is decidedly intellectualist in that I argue that no skilled action can be performed without the recruitment of executive functions. This implies that no skilled action can be performed without a minimum amount of top-down cognitive effort.


    Amaya, S. (2013). “Slips.” Noûs 47:3, 559–576.

    Baddeley, A. (2007). Working memory, thought, and action. Oxford University Press.

    Diamond, A. (2013). “Executive functions.” Annual Review of Psychology 64, 135–168.

    Evans, J. St. B. (2010a). “Intuition and reasoning: A dual-process perspective.” Psychological Inquiry 21:4, 313–326.

    Evans, J. St. B. (2010). Thinking twice: Two minds in one brain. Oxford University Press.

    Iaccarino, L., Chieffi, S., & Iavarone, A. (2014). “Utilization behavior: What is known and what has to be known?.” Behavioural Neurology 2014, 1–9.

    L’Hermitte, F. (1983). “`Utilization behaviour’ and its relation to lesions of the frontal lobes.” Brain 106, 237–255.

    Rietveld, E. & Paglieri, F. (2012). “Bodily intentionality and social affordances in context.” Consciousness in interaction. The role of the natural and social context in shaping consciousness, 207–226.

  7. Hi Juan Pablo,

    Thanks for the very interesting paper!

    Like Josh, I take Anscombe and Anscombeans to hold that knowledge of what you are doing is somehow present in the action itself, and is not the result of a separate act of attending to or reflecting upon the action. (As I’m thinking of the Anscombean view, it might for instance involve demonstrative ways of thinking about actions that latch onto their referents via the corresponding lower-level motor capacities.)

    And I was wondering whether Anscombeans of this stripe might have a different response to the expert confabulation cases you describe. In general, if you have two ways of finding out about a certain subject matter, it is not surprising if they give divergent results on occasion. So, perhaps, the confabulating expert knows what she is doing in the Anscombean in-action sort of way, while she fails to know in the other, reflective sort of way.

    Something similar could apply to the amnesia cases as well: the agent knows what she is doing as she’s doing it, but since she is not reflectively attending upon her own actions she does not commit those details to memory.

    What do you think?

  8. Hi Markos,

    Thanks for your comments! I’ve posted my thoughts on them in the thread where Josh and Michael have been posting.

  9. Hi Markos!

    You write:

    As I’m thinking of the Anscombean view, it might for instance involve demonstrative ways of thinking about actions that latch onto their referents via the corresponding lower-level motor capacities.

    I’m wondering if this way of cashing out practical knowledge will have the resources to distinguish between skill and luck. That is, would my succeeding at a task be sufficient for demonstrating that I’m latching onto the right referents via lower-level motor capacities? What about if I’m just lucky and latch onto the right things but under the wrong descriptions or accidentally refer to the relevant properties with the right lower-level motor capacities? Would these be possible on the view you are suggesting?

    Thanks ,

Comments are closed.