Guidance of Visual Attention

28 thoughts on “Guidance of Visual Attention”

1. Introduction

   First, thanks to the conference organizers for their invitation to comment, and to Denis Buehler for his thought-provoking paper!

   The paper’s main goal is to use a concept deployed in psychological explanations (“the central executive”) to respond to a philosophical challenge raised by Harry Frankfurt: to explicate the distinction between processes that are guided or unguided by an individual (p. 1). As such, the paper is an excellent example of the fruitful kinds of engagement that can occur between psychology and philosophy.

   Here is a rough outline of the paper. First, Buehler argues that the “primary realizer” of an agent’s guidance of her shifts of visual attention is her central executive system’s control of those shifts (p. 13). He goes on to argue that his thesis about the realization of guidance helps illuminate the concept of guidance (p. 18).

   His paper raises many issues that are worth discussion. In this commentary, I’ll focus on just one. Here’s a brief preview of my worry. Because Buehler wants to use the concept of the central executive to illuminate the concept of guidance, he seems to require that we understand the concept of the central executive better than we understand the concept of guidance. But I’m not sure that we do.

   In the rest of this commentary I’ll expand on the worry, and then end with a constructive suggestion about how it might be addressed.

   1. Buehler on the central executive

   On Buehler’s view, all we can say about the central executive for now is that “the central executive system organizes and coordinates central processing and storage of resources for the execution of cognitive processes, including individuals’ goal-represented activities” (p. 14). He also identifies at least three functions carried out by this system: switching from one task to another, maintaining task-relevant memories, and inhibiting certain kinds of responses. (pp. 14-15). He says that his goal in characterizing the central executive in this way is to abstract away from the details of any specific model of the central executive, since it would be hasty to commit to the success of any particular model.

   Buehler thinks that this concept will help illuminate the concept of guidance. As he says, “[appeals] to the central executive system help us understand individuals’ guidance by connecting it with a kind – central executive control – that appears in psychological explanations” (p. 20). On this way of thinking, we can look to psychological explanations that refer to “central executive control” to learn something about guidance (some of the instances of it, anyways – guidance in creatures similar enough to actual primates).

   2. Baddeley on the central executive

   In order to make my worry about the extent to which we understand the central executive clear, I need to briefly discuss Baddeley and Hitch’s original model of the central executive.

   In 1974, Baddeley and Hitch proposed a model of working memory with three components – a visuo-spatial sketchpad, a phonological loop, and a central executive. A great deal of the current usage of the term “central executive” can, I think, be traced to their use of this term. Here is how Baddeley, in a more recent publication, has described his conception of the central executive:

   “The CE [central executive] is the most complex component of WM [working memory]. Within the original model it was assumed to be capable of attentional focus, storage, and decision making, virtually a homunculus, a little man in the head, capable of doing all the clever things that were outside the competence of the two subsystems.” (Baddeley 2012 pp.13-14)

   Baddeley refers to his conception of the central executive as homuncular; i.e., not really explanatory. But he is aware of this apparent problem, and has a reply to it:

   “Although our model tended to be criticized for taking this approach… I regard homunculi as potentially useful if used appropriately. It is important that they are not seen as providing an explanation, but rather as a marker of issues requiring explanation. Provided the various jobs performed by the homunculus are identified, they can be tackled one at a time, hopefully in due course allowing the homunculus to be pensioned off.” (Baddeley 2012 p.14)

   So on Baddeley’s approach, the term “central executive” shouldn’t actually show up in any complete psychological explanation. Instead, it’s the job of some psychological explanations to explain various aspects of the central executive, and eventually, collectively, to give us reason to do away with the concept. One psychologist – Logie – has even recently argued that we’ve arrived at the time to do away with the concept in his recent paper “Retiring the central executive” (2016).

   Miyake et al., who Buehler cites in footnote 25 as a source on the central executive, seem to approach the central executive like Baddeley. They say that their paper is meant to address the problem that “the field still lacks a compelling theory of executive functions – general purpose control mechanisms…” (Miyake et al. 2000 p. 50). Therefore, they seem to think that the central executive is something to be explained rather than something can be used to explain other phenomena.

   This seems like a fine way for psychologists to understand the central executive system. However, it seems to me that this approach to the central executive is in tension with Buehler’s project. Buehler wants to use the central executive to explain guidance. But if the central executive is itself something that is in need of explanation, it seems to me that this explanatory strategy might not be successful. It we want to explain guidance, then we should advert to something which does not itself require explanation – and especially not something that psychologists intend to eventually explain away. Instead, we should advert to something already well understood, or something that we have good reason to think will be easier to explain than guidance.

   3. Moving Forward

   I’d like to finish things off by outlining, in a very rough form, one possible method of addressing the problem I’ve raised. It’s modelled off of Wayne Wu’s (2014) approach to perceptual attention.

   To put it quickly, Wu thinks that despite apparent disagreements between psychologists about what perceptual attention “really is”, an analysis of the experimental practice of many psychologists shows that there is an account of perceptual attention that they tacitly agree on (Wu 2014 pp. 38-40).

   On this kind of approach, we don’t learn about perceptual attention just by looking at what is said by psychological explanations that deploy the term “perceptual attention”. Instead, we look at those explanations, but we should also keep in mind how well they cohere with the best account of what many psychologists really mean when they say “perceptual attention”. This kind of approach is empirically informed, but critical and careful in a particular respect.

   Perhaps analogous work on the term “central executive” would help Buehler address the worry I’ve raised here. It might be that the best analysis of what certain psychologists have to say about the central executive gets us the result that the central executive is better understood than Baddeley thinks it is, or at least not a concept that should eventually be done away with.

   References

   Baddeley, A.D. (2012) “Working Memory: Theories, Models, and Controversies”. Annu. Rev. Psychol. 63:1–29
   Baddeley, A.D. & Hitch, G.J. (1974). Working memory. In G.A. Bower (Ed.), Psychology of learning and motivation (Vol. 8, pp. 47–90). New York: Academic Press.
   Logie, R. H. (2016). “Retiring the central executive”. Q J Exp Psychol. 19:1-17.
   Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., Howerter, A., & Wager, T. D. (2000). The unity and diversity of executive functions and their contributions to complex “frontal lobe” tasks: A latent variable analysis. Cognitive Psychology, 41: 49–100.
   Wu, W. (2014). Attention. New York: Routledge.

   Loading...

2. Introduction

   I am grateful to the organisers of the Minds Online conference for inviting me to comment on Denis Buehler’s excellent paper. Buehler contributes to action theory by developing an empirical explication of individuals’ guidance. The specific empirical case study he focuses on is the psychological account of how shifts of attention in visual search are controlled by cognitive mechanisms. He concludes that the controlling mechanism that realises individuals’ guidance of shifts of attention in primates is the central executive system. From this, Buehler argues for the more general claim that the “control of a process by an individual’s central structures (partly) constitutes that individual’s guidance” (Section 1, Paragraph 11).

   I am sympathetic to the general constitutive claim Buehler offers, and agree that controlling capacities of an organism’s central structures play an important role in how the organism guides its behaviour. However, I have worries with regard to the more specific claim—concerning the relations between shifts of visual attention, central executive functions, and guidance—that Buehler builds upon to reach his conclusion, and think that Bueller has chosen a case study that does not serve his purpose. More specifically, I am not convinced by the arguments offered in the paper to the conclusion that central executive control of shifts of attention realises individuals’ guidance, since I am not convinced that a shift of attention is a guided behaviour in a relevant sense of the term.

   Shifts of attention and its controlling mechanisms

   Before introducing my worries in detail, I would like to amend Buehler’s characterisation of what psychology and neurobiology have uncovered about shifts of attention and their controlling mechanisms. Buehler purposefully abstracts away from specific computational and neural mechanisms, however, as I shall argue below, there are crucial details that seem to be in tension with his conclusion regarding attention shifts, executive functions and guidance.

   Attention comes in many different varieties, like overt or covert; focal or residual; spatial, feature-based, or object-based; endogenous or exogenous, etc. (Chun et al., 2011; Carrasco, 2011). The distinction that is relevant from the perspective of Buehler’s project is the endogenous versus exogenous distinction, since while the other terms characterise where and how attention is allocated, this distinction characterises how the allocation of attention is initiated and controlled. Endogenous attention is activated in a top-down manner by internal causes like goals, beliefs and desires, and is typically under voluntary control, whereas exogenous attention is stimulus-driven, bottom-up, involuntary and automatic. (Note, however, that Wu (2014) claims that the top-down versus bottom-up and the controlled versus automatic distinctions come apart, and Jennings (2012) argues that the endogenous versus exogenous and voluntary versus involuntary distinctions are different).

   As it has recently been argued, endogenous and exogenous attention correspond to distinct underlying networks in the brain. The functioning of attention in general is associated with the functioning of three different networks: an alerting network that maintains optimal vigilance during tasks, an executive network that is responsible for target detection for focal processing, and an orienting network that prioritises sensory input (Posner and Petersen 1990; Petersen and Posner 2012; to see how these networks are related to executive functions see e.g. Diamond, 2013).

   The so-called orienting network consists of two partly overlapping networks that can be corresponded to top-down and bottom-up orienting. The activity of the so-called dorsal frontoparietal network, i.e. frontal and parietal areas along the dorsal stream (like the intraparietal sulcus and the frontal eye field), influences perceptual processing in accordance with actual task demands, and is responsible for goal-directed stimulus-response selection. The other, more lateralised, so-called ventral frontoparietal network (including the temporoparietal junction and the ventral frontal cortex) plays a circuit-breaking role by breaking ongoing cognitive activity when a salient unexpected stimulus is detected. On the basis of these functions, the dorsal system is associated with orienting endogenous attention, whereas the ventral system is associated with orienting exogenous attention (Corbetta and Shulman, 2002; Petersen and Posner 2012; Wu 2014).

   Although Buehler claims that the “exogenous system exclusively responds to stimuli that are physically salient” (Section 3.1.1, Paragraph 1), there is empirical evidence to the contrary: the ventral frontoparietal network is sensitive to behaviourally relevant stimuli, i.e. it plays its circuit-breaking role if behaviourally relevant stimuli are detected outside focal processing. Corbetta and Shulman (2002) propose that information about behavioural relevance is mediated by connections between the ventral and the dorsal systems. That is exogenous attention is partly driven by goals, beliefs and task demands representing areas that are typically associated with top-down, voluntary, goal-directed control. Even more importantly from our present perspective, there are tight interactions between top-down and bottom-up processing in the dorsal frontoparietal system as well. Corbetta and Shulman (20002) report that the dorsal network is modulated not only by contextual, task-relevant information but also by the sensory distinctiveness of stimuli, i.e. by information that is typically associated with bottom-up, automatic, stimulus-driven control (Corbetta and Shulman, 2002).

   The existence of such interactions between top-down and bottom-up control of attention is supported by state-of-the art computational models of neural effects of attention as well. According to the so-called Normalisation Model of Attention (Reynolds and Heeger, 2009; Lee and Maunsell, 2009), neural output is jointly determined by three distinct factors: stimulus features, the operation of an ‘attentional field’—that exerts its effects by multiplying the excitatory input of the neurons in question,—and built-in input processing mechanisms that normalise neural responses via divisive suppression, i.e. by dividing the stimulus drive by the activity of a pool of surrounding neurons (see Herrmann et al. 2010, and Cutrone et al., 2014 for empirical support; Beuth & Hamker, 2015 for a mechanistic microcircuit model; and Wu, forthcoming for a philosophical summary). Crucially from our present perspective, within this model the ‘attentional field’ is jointly determined by long range frontal back-projections mediating top-down information about task relevance and bottom-up signals mediating the effects of attentional capture (Heeger, personal communication).

   The output of these neural computations (amplified and/or sharpened neural representations)—and thus ultimately the tight interaction between top-down, task-related, and bottom-up, sensory control mechanisms (see also Corbetta and Shulman, 2002)—generates the so-called saliency maps (Itti and Koch, 2000, 2001) that prioritise stimulus-representations on the basis of their overall salience, serve as the basis of serial selection for recognition and action, and thus play a crucial role in directing shifts of attention during visual search tasks.

   Central executive control and individuals’ guidance

   With all this at hand, consider Buehler’s argument for the empirical explication of individuals’s guidance in terms of central executive control. Buehler acknowledges that for central executive control to help explicate individuals’ guidance central executive control and individuals’ guidance must appropriately correlate.

   Given the distinction between central executive (i.e. endogenous) and exogenous control of shifts of attention, Buehler needs to establish that individuals’s guidance appropriately correlates with endogenous control and not with exogenous control. From this perspective, the tightly intertwined nature of endogenous and exogenous control (see previous section) poses a problem. How can individuals’ guidance correlate with one but not the other?

   Buehler argues that the crucial difference is that while goal-directed endogenous control drives attention towards the target, stimulus-feature based exogenous control can drive attention away from the target and thus interfere with goal directed search. This claim, however, does not respect the actual mechanism underlying shifts of attention. Recall that under realistic conditions in visual search tasks there are no simple cases of purely endogenously or exogenously driven shifts of attention. In the case of what is traditionally conceived of as an endogenous shift of attention, certain perceptual representations of specific stimulus-features throughout the visual field or at specific spatial locations (determined by actual task demand) are enhanced, which modulates the actual saliency map. Nevertheless, attention will be shifted to the stimulus with the highest saliency, no matter which stimuli were targeted during the task. So in cases where top-down attentional amplification is not able to push the targeted stimulus to the top of the priority list of the saliency map attentional shifts will wander from distractor to distractor until finally it lands on the target. Note that although such a case is often called an instance of goal-directed allocation of attention, strictly speaking attention is not directed towards the target—it is directed towards the stimulus with the highest salience, which might or might not be the target. Similarly, just as endogenous control does not shift attention ‘towards’ the target, exogenous control does not shift attention ‘away’ from the target. Put it simply, shift of attention is always captured by the stimulus with the highest salience.

   That is, if one thinks that such endogenously controlled shifts of attention are guided, then it is hard to see why exogenously controlled shifts would not count as guided. And vice versa: if one thinks that exogenously controlled shifts of attention are not guided, it is hard to see why endogenously controlled shifts would count as guided. From this perspective, the description in the previous paragraph suggests that neither exogenous nor endogenously controlled shifts of attention are guided in the relevant sense.

   To see why, consider how individuals’ guidance might be characterised at the pre-theoretical level. Buehler offers an analogy with a mountain guide. According to Buehler, a good guide “picks the peak as her expedition’s destination and orients her expedition toward the peak”, “knows the route to the peak”, and “stably steer toward the goal of the expedition” (Section 3.2, Paragraph 2). Compare this characterisation with the description of how shifts of attention are determined. We have already seen the major problem: shifts of attention, even endogenously controlled shifts of attention do not know the ‘route to’ the target, they do not ‘stably steer toward’ the target. If central executive control of shifts of attention is like a mountain guide, then it is a pretty bad one. Imagine a mountain guide who only enhances the chance of reaching the summit the expedition wants to reach, who checks if they have reached the right summit only after the expedition has gotten onto the top, and if not then moves the expedition on towards a next summit which might or might not be the one they want to reach. Shifts of attention are guided just as much as such an expedition is.

   The situation is not better even if instead of Buehler’s mountain guide analogy, we rely on Frankfurt’s (1978) original characterisation of individuals’ guidance. As Frankfurt puts it, a behaviour is guided in the relevant sense if “its course is subject to adjustments which compensate for the effects of forces which would otherwise interfere with the course of the behaviour”, such that these compensatory adjustments tend to “ensure that the behavior is accomplished” (Frankfurt, 1978, p. 160). However, as we have seen it, central executive control of shifts of attention in visual search is not able to exert such adjustments that could compensate for the effects of salient distractors, and it cannot ensure that the visual search gets accomplished. Therefore, endogenously controlled shifts of attention do not seem to be guided in the original sense of the term.

   Conclusion

   In this brief comment I argued that endogenously controlled attention shifts in visual search are not forms of guided behaviour in the sense relevant for action theory, and thus central executive control of shifts of attention cannot realise individuals’ guidance. This means that Buehler’s present argument is not sound, since its first premise is false. However, it does not seriously jeopardise Buehler’s overall project. Buehler could still establish his general constitutive claim (“the control of a process by an individual’s central structures partly constitutes that individual’s guidance”) by showing that whenever a behaviour is guided it is under the control of some central structures.

   References

   Beuth, F., and Hamker, F. H. (2015): A mechanistic cortical microcircuit of attention for amplification, normalization and suppression. Vision Research, 116: 241-257.
   Carrasco, M. (2011): Visual attention: The past 25 years. Vision Research, 51: 1484-1525.
   Chun, M, Golomb, J D, and Turk-Browne, N. (2011): A Taxonomy of External and Internal Attention, Annual Review of Psychology, 62: 73-101.
   Corbetta, M., Shulman, G. L. (2002): Control of goal-directed and stimulus-driven attention in the brain. Nature Review Neuroscience, 3: 201–215.
   Cutrone, E. K., Heeger, D. J., and Carrasco, M. (2014): Attention enhances contrast appearance via increased input baseline of neural responses. Journal of Vision, 14(16): 1-14.
   Diamond, A. (2013): Executive functions. Annual Review of Psychology, 64: 135-168.
   Frankfurt, H. G. (1978): The problem of action. American Philosophical Quarterly, 15: 157-162.
   Herrmann, K., Montaser-Kouhsari, L., Carrasco, M., and Heeger, D. J. (2010): When size matters: Attention affects performance by contrast or response gain. Nature Neuroscience, 13(12): 1554–1559.
   Itti, L., and Koch, C. (2000): A saliency-based search mechanism for overt and covert shifts of visual attention. Vision Research, 40: 1489-1506.
   Itti, L., and Koch, C. (2001): Computational modelling of visual attention. Nature Review Neuroscience, 2: 194-202.
   Jennings, C. D. (2012): The Subject of Attention. Synthese, 189 (3): 535-554.
   Lee, J., and Maunsell, J. H. R. (2009): A normalization model of attentional modulation of single unit responses. PLoS ONE, 4(2): e4651.
   Petersen, S. E., Posner, M. I. (2012): The attention system of the human brain: 20 years after. Annual Review of Neuroscience, 35: 73-89.
   Posner, M. I., Petersen, S. E. (1990): The attention system of the human brain. Annual Review of Neuroscience, 13: 25-42.
   Reynolds, J. H. and Heeger, D. J. (2009): The normalization model of attention. Neuron, 61(2): 168–185.
   Wu, W. (2014): Attention. New York: Routledge.
   Wu, W. (forthcoming): Shaking up the mind’s ground floor: the cognitive penetration of visual attention. Journal of Philosophy.

   Loading...

3. I am very happy to comment on Denis Buehler’s interesting paper ‘Guidance of Visual Attention’. I will start with a brief summary of how I understand Denis’ goal and argumentative strategy, and then move on to comments and critical remarks.

   Summary

   Denis’ paper aims to contribute to a solution of Harry Frankfurt’s problem of action, which – in Frankfurt’s own words – is to “explicate the contrast between what an agent does and what merely happens to him” (Frankfurt 1988 [1969], p. 69). Following a central idea in Frankfurt’s work, Denis starts from the assumption that the relevant contrast is between processes that are guided by the relevant individual and those that are not under the individual’s guidance (p. 1). After trying to show that some of the most prominent contemporary accounts of agency fail, since they either fail to be comprehensive enough to cover simple and non-human agency in animals, or fail to explain why guidance is attributable to the whole individual, Denis then aims to explicate the relevant concept of guidance by means of studying a central case, namely the individual’s guidance of visual attention. In this case, Denis argues, guidance is realized by the central executive control system. We can explicate the notion of guidance by appeal to central executive control, he argues, because (a) we can empirically show that such control correlates with an individual’s guidance (p. 17), and (b) appeal to such control deepens our understanding of guidance and thus helps to ground explanations involving the notion of guidance (p. 2, 17). On the basis of this case study, Denis finally conjectures that, generally, an individual’s guidance is at least partially constituted by “[t]he control of a process by an individual’s central structures” (p. 24).

   I really like what Denis takes as his case study. In the case of attention shifts the distinction between active and passive has been studied empirically in great detail, and so I share the hope that a close look at this case study may help to make progress with the problem of action. The case is also interesting for two other reasons. First, consider that visual attention shifts are clearly mental acts that are not “rehearsals of physical actions” (Carruthers 2009, p. 152) and do not involve any “visual and other forms of imagery” (ibid.). Anyone who, like Carruthers, still thought that there really aren’t mental acts thus is shown to be mistaken. Second, consider that if some stretches of attention are cases of agency that are under the individual’s guidance then it is hard to see how attention could be a necessary part of the explanation of individually guided agency since we would already need attention in order to get attention.

   Strategy and Overall Shape of the Project

   Let me now move to some comments on the general shape of the project. Denis says that he is interested in an explication of the concept of an individual’s guidance. While he clearly wants such explication of a concept to be different from conceptual analysis (p. 2), I was wondering why Denis talks about concepts at all. Why not just talk about providing an account of what constitutes an individual’s guidance, i.e. at least a partial account of (the nature of) guidance? Given that Denis clearly believes that empirical work can play a central role in his project, isn’t that project better captured by dropping talk of concepts (physics tells us about the nature of space not the concept of space, biology about the nature of tigers not the concept of a tiger, and the psychology of attention about the nature of guidance and not the concept of guidance)? While I don’t think it matters much for what follows, I would like to invite Denis to comment on why he frames his project in terms of conceptual explication.

   The Critique of Recent Action Theory

   Denis argues that many recent theories of agency violate the comprehensiveness constraint: they are focused on sophisticated human agency and stand no hope to explicate what distinguishes the active from the passive movements of a spider. I tend to agree with this complaint. But the complaint rests on the assumption that these accounts are trying to explicate the same phenomenon that Denis is trying to explicate. An opponent might here disagree on what the relevant explicandum is. She might think that the most significant kind in the relevant neighborhood (or maybe the most important concept) is the sophisticated, characteristically human kind. If the objector were able to show that this “sophisticated agency” has a powerful explanatory role, normative structure, and/or central place in our conceptual framework that is not shared by primitive agency, then – I believe – we cannot fault them for not accounting for primitive agency or primitive guidance. An opponent of this type might agree that there we need an account of this primitive agency, but this would not solve or even contribute much to solving the philosophical problem of action. In order to counter this objection, I believe, one would need to say a bit more about what makes primitive agency significant.

   Routine Action. Guidance or Control?

   I was wondering about how Denis understood the notion of guidance that he sought to explicate. Harry Frankfurt, correctly I believe, allowed for cases of agency where a process is under the individual’s guidance even though she is not currently controlling the process in any sense (p. 75). He uses the example of driving downhill. You might actively drive (and guide the driving) even at times when you do not intervene to adjust the movement that his happening though gravitational forces. In cases like this the individual’s guidance consists in her preparedness to intervene if necessary, and her being in a position to do so more or less effectively (ibid.). Many routine actions arguably are similar: walking toward the kitchen in the morning, effortless reading, or playing a well-rehearsed piece on an instrument are an individual’s acts that she doesn’t control. They are under her guidance in Frankfurt’s sense, since she could effectively intervene if necessary, but she does not currently control them. I wasn’t sure that Denis does and can allow for such cases of guidance. This issue comes up in two places.

   First, there is Denis’ critique of accounts of agency that fail to distinguish guidance from background conditions of guidance. Why couldn’t an intention based account, for example, conceive of intentions as types of states whose functional role is to keep the agent “on track” (and whose existence explains why she is able to adjust and interfere if necessary). Denis seems to want guidance to be more “online” than this. But I am not sure that one can do this while accounting for the driving type cases.

   Second, the issue of routine action comes up with regard to Denis own account. It is at least unclear that cases like this require the central executive or executive function. Executive function is normally thought to be required exactly when actions have not been routinized. When they are, they do not require executive resources. This problem for Denis’ approach can be put as a dilemma. If guidance by the individual requires her current causal control, then many forms of agency are not guided by the individual and thus the category of processes that are guided by the individual is not even approximately coextensive with the category of ‘what an agent does’. But if guidance is understood more broadly (as Frankfurt understands it), then it is highly unclear that the central executive system is necessary for guidance as in routine actions. We would thus like to hear more from Denis about the central executive as he thinks of it that makes room for its deployment also in routine action.

   What is the central executive system? Does it explain anything?

   Denis, importantly, notes that recent research shows that the distinction between active and passive attention shifts does not coincide with the distinction between endogenously and exogenously generated attention shifts. Past experience and reward, for example may initiate a passive attention shift. The passive attention shift is endogenously generated and controlled. Which endogenous attention shifts then are guided by the individual? Denis thinks that it’s those that are controlled by the central executive. I worry about how explanatory this suggestion is. The concept of the central executive system, as I know Denis is aware, was first introduced by Alan Baddeley in his highly influential work from the 1970s as an important component of working memory. In a recent review of work that was influenced by his research Baddeley notes that the invocation of the central executive has often been criticized for invoking a sort homunculus that can do all the things the subsystems cannot do. For this reason, Baddeley says, appeal to the central executive itself should not be understood “as providing an explanation, but rather as a marker of issues requiring explanation (Baddeley 2011, p. 14). As Denis notes there are a number of functions attributed to the central executive: task switching, working memory maintenance, and inhibition of distractors. Not all of these are involved in every active attention shift (or visual search). But what then is it supposed to mean that “the central executive” controls the attention shift? Denis says various things about that: on p. 19, for example, he says that “the central executive system functions to integrate and coordinate individuals’ states and competencies for the successful execution of individual-level processes”. What does the explanatory work, it seems to me, here isn’t a quasi-homuncular executive system, but the fact that the individual’s states and competencies are coordinated and integrated so as to allow the successful pursuit of her goal. The individual guides her activity, because it isn’t the case that one of her subsystems is controlling the act by itself, but rather she can bring to bear a wide variety of states and competencies in pursuit of her goal. My question for Denis then is: is guidance really explained by “control by the central executive” or rather by the integration and coordination of a wide variety of individual level states, processes, skills and competencies? On the alternative picture then there isn’t a “central structure” (p. 25) that coordinates the parts of the individual, but there is just the integration and coordination of the parts and states of the individual.

   Is there an a priori argument for Denis’ thesis?

   Finally, I would like to ask about the role of the empirical research in the explication of guidance. Step back for a moment, and consider complex objects generally. Which states and processes are fundamentally states and processes of the whole object, and which are fundamentally states and processes of its parts? Consider the molecular structure of some complex molecule. That structure clearly is a property of the whole complex object and not its parts because having that structure is a property that involves relations between the parts (no single part, or sub-system, has that structure). Generally, it seems very plausible that a property is fundamentally a property of a whole W with parts pp and not any of its parts p if having that property requires that those pp‘s to stand in specific (external) relations. But then it seems to follow more or less a priori that a process is guided by the individual if it is guided in a way that integrates and coordinates its various sub-systems (of course, there is a question of who “big” the relevant whole really is, since it is unlikely that all subsystems of the individual are integrated and coordinated). So, I would like to invite Denis to say more about what exactly we are supposed to gain from the empirical data he appeals to.

   Literature

   Baddeley, A. (2012). Working Memory: Theories, Models, and Controversies, Annu. Rev. Psychol., 63, 1–29
   Carruthers, P. (2009). Action-Awareness and the Active Mind, Philosophical Papers, 38(2), 133-156
   Frankfurt, H. (1988 [1978]). The problem of action, Reprinted In: H. Frankfurt. The importance of what we care about. Philosophical essays (Ch. 6, p. 69-79). Cambridge, UK: Cambridge University Press

   Loading...

4. I want to thank Peter, Mark, and Sebastian for taking the time to read, think about, and comment on my paper. Having people spend their time on one’s own work is a humbling experience, especially for someone early in their career. I am deeply grateful for it. I’ve learned a lot from working through the three sets of comments. And I hope I am doing justice to each of them.

    

   Reply to Fazekas: Captured attention

   I really enjoyed the extent to which Peter delved into psychological research relevant to my paper. Peter thinks that I misuse or misunderstand our psychological evidence concerning visual attention shifts. Naturally, I disagree. Let me try to explain why.

   A loud bang, or bright light flashed at you, will capture your attention. There is some reason to think that you may be able to voluntarily suppress such a shift, if you really try, or if you focus your attention on some other thing. But, typically, such a sudden, abrupt, physically salient stimulus will capture your attention. In the paper, I identify this type of episode as capture of attention. I claim that such shifts of attention are intuitively passive, non-guided episodes. Psychological processes underlying this type of episode hence do not appropriately correlate with individuals’ guided shifts of attention.

   In the paper, I report that psychologists explain such captured attention shifts in terms of the exogenous system’s activity alone. I write: “the exogenous system exclusively responds to stimuli that are physically salient (p. 10).” Peter mistakenly thinks that I claim that the exogenous system does never respond to stimuli other than physically salient ones. This misunderstanding is my fault. I should have written “the exogenous system alone exclusively …” If I were to make the claim Peter takes me to be making, I would be in contradiction with what I say in the next paragraph, where I address ways of interaction between the exogenous and the endogenous system, and mention research to the effect that endogenous goals can “tune the exogenous system to be more sensitive” to certain stimuli.

    

   Does Peter want to claim that physically salient stimuli alone never determine where attention shifts? That claim seems antecedently implausible – remember the above examples of captured attention. The claim is also not supported by psychology.[i] Even the authors that Peter cites seem to acknowledge such captured attention shifts.[ii]  Recent computational models do not support the claim.[iii] Capture as in the example above will not be mediated by your current expectations or goals. The behavioral relevance of capturing stimuli lies in the evolutionary advantage that comes with keeping track of abrupt changes in your environment. We must distinguish such cases from the influence that current (endogenous) goals have on attention-shifts when they, for instance, guide attention toward a tool relevant for an ongoing bodily action.[iv] Peter is right to point out that most of the time, attention shifts are influenced by our occurrent goals or other endogenous factors. And indeed, psychologists sometimes have called some of these types of shifts instances of ‘captured’ attention. But I find this nomenclature misleading – after all, the individuals’ goals drive these attention shifts. In the paper I focus on what I take to be clear cases of capture.[v]

    

   Peter writes:

   “Given the distinction between central executive (i.e. endogenous) [my emphasis] and exogenous control of shifts of attention, Buehler needs to establish that individuals’ guidance appropriately correlates with endogenous control and not with exogenous control. From this perspective, the tightly intertwined nature of endogenous and exogenous control … poses a problem. How can individuals’ guidance correlate with one but not the other?”

   The passage suggests that Peter takes himself to have established that endogenous and exogenous activity are always ‘entwined.’ (Or else there would be no in-principle problem with the approach Peter worries about.) I have explained why I think that Peter has not established this.

   As I acknowledged earlier, Peter is certainly right to think that endogenous and exogenous factors interact, most of the time, so as to generate shifts of attention. In section 3.1 I explicitly acknowledge the interaction between exogenous and different types of endogenous factors. But in the quote above, Peter mistakenly suggests that I want to explicate guidance as control by endogenous factors, and in contrast with exogenous factors. The point of section 3.1 is rather to distinguish which endogenous factor plausibly correlates with individuals’ guidance, and which endogenous factors do not.[vi]

    

   So which factor does correlate with individuals’ guidance? Peter claims that

   “Buehler argues that the crucial difference is that while goal-directed endogenous control drives attention towards the target, stimulus-feature based exogenous control can drive attention away from the target and thus interfere with goal directed search,” and that my answer to the question “does not respect the actual mechanism underlying shifts of attention.”

   This portrayal is not quite accurate. I believe that, first, there is a sense in which the relevant endogenous factor drives search toward the search goal, but, second, both endogenous and exogenous factors can contribute to driving attention away from the actual target.

   Computational models and behavioral studies of visual search suggest this more complex picture. During visual search, a representation of the search goal held in memory determines the computations that lead to assignments on the priority map. Our best models of shifts during search assume that attention moves to the location on the map that has the highest priority value. This value is determined, partly, by the similarity between the represented goal and stimuli at specific locations. So these models offer a concrete sense in which the search goal causally controls the computations of assignments on the priority map (or “drives” attention toward the goal).

   Other factors, too, figure into the determination of the priority value for any given location on the map. These factors include the physical saliency of a location, information about types of scenes, or priming for specific features. Insofar as these influences derive from factors other than the search goal, they draw attention away from its goal.[vii] But even when attention lands in the wrong location, we can say that the processes are ‘guided toward some search goal.’[viii]

   How can we distinguish the search goal’s influence on the priority map from that of other endogenous factors, such as other memories in working memory? Behavioral studies suggest that the difference is due to the allocation of central resources by a central executive system. My proposal is that, in spite of the constant interaction between the different factors, appeal to psychology helps us identify one endogenous factor – central executive control – that does correlate with individuals’ guidance.

    

   Finally, how can the executive system’s control of attention shifts help us illuminate marks of the concept guidance, such as ‘knowing the route to the goal’ or ‘stably steering toward the goal’? Illumination comes from appeal to the more sharply characterized scientific notions. Naturally, we cannot expect the system to do its job more or less without fail. After all, the system does not know where the search target is located. Still, it has a template of its search target and uses surprisingly sophisticated and successful computations to search for that goal as efficiently as possible.[ix] And the system won’t let off until it has actually found its target. The system’s situation, in visual search, is more like that of a mountain guide who merely owns a drawing of the peak’s contours, and is supposed to navigate the expedition toward that peak in a foggy, cluttered environment. Given these conditions, success rates like the attentional system’s are impressive. I’d follow that mountain guide on her expedition. Well, if I absolutely had to go on the expedition, at any rate.

    

   Reply to Fortney: Explanation in terms of central executive control

   Mark worries that I want to

   “use the concept of the central executive to illuminate the concept of guidance, [and hence that I] … require that we understand the concept of the central executive better than we understand the concept of guidance. But [Mark is] not sure that we do.”

   Therefore, I presume, appeal to central executive control does not illuminate guidance. The worry is natural. Philosophers and psychologists alike have criticized research on executive functions as not explanatory. I believe that these criticisms are now out-dated. Unfortunately, I did not have space, in the paper, to expand on my views about research on executive functions. That is why I am particularly happy that Mark voices his concerns about the issue.[x]

    

   Baddeley writes of the central executive system (cf. Mark’s quote):

   “Within the original model [the central executive] was assumed to be capable of attentional focus, storage, and decision making, virtually a homunculus.”

   In Baddeley’s original model the central executive was homuncular, because the very abilities that psychologists wanted to explain were attributed to the system. Psychologists wanted to explain psychological competencies of individuals, such as decision-making. Ideally they wanted a neuro-computational mechanism that would explain such processes. Attributing an ability to decide to a system, of course, does not explain individuals’ ability to decide. And it is far from a fully mechanistic explanation of decision-making. A series of psychologists and philosophers, accordingly, considered the central executive system to be merely a placeholder.[xi]

    

   Notice the following passage in Mark’s second quote from Baddeley, however:

   “Provided the various jobs performed by the homunculus are identified, they can be tackled one at a time, hopefully in due course allowing the homunculus to be pensioned off.”

   Psychological explanation typically advances from identifying an explanandum through a series of stages of ‘breaking-down’ the explanandum into components. Each movement from a higher-level to a lower-level stage constitutes explanatory progress. A psychological process such as object-vision is identified as plausibly involving the sub-components of image-based and surface-based processing. Next, psychologists tackle the problem of explaining each sub-component.

   The psychological investigation of executive functions has seen progress along these lines, in the last 20 years. I believe that especially Miyake et al.’s introduction of confirmatory factor analysis into this field was an important step.[xii] It allowed developing and testing hypotheses about the different sub-components that characterize the central executive system. In this way, Miyake et al. identified three plausibly basic executive functions: switching, inhibition, and maintenance. This work allowed breaking these sub-components down into further sub-components.[xiii] The process of breaking down functions allowed more precise behavioural study of the components, developing computational models of the components, and identifying neural structures that may implement the different components.[xiv] It allowed investigating the role of executive functions in more complex cognitive tasks, and their relations to such issues as general intelligence, ageing, anxiety, and others.[xv] Many recent developments in the field constitute genuine explanatory progress.

    

   How, then, do appeals to central executive control deepen our understanding of individuals’ guidance? The characterization of the system may be preliminary and will certainly develop with future research. Appeals to the system nevertheless deepen understanding of guidance in at least two ways. A first way in which such appeals deepen understanding is by connecting our pre-theoretic notion with notions that psychologists investigate. Such a connection can help sharpen the pre-theoretic notion. There can be no doubt that current psychological research provides characterizations of executive functions that are sharper than our pre-theoretic notion guidance. A second way is that, thereby, we can see that the knowledge psychologists have acquired or will acquire about the system – about computational models of executive functions, behavioural profiles, their connection with general intelligence and so forth – in principle bears on and can help illuminate our pre-theoretic notion. That knowledge was not part of our pre-theoretic understanding of guidance.

    

   I think that Mark is right to caution against identifying the central executive system “just by looking at what is said by psychological explanations that deploy the term “[central executive control]”.” More generally, Mark seems right when he seconds Wayne Wu’s recommendation that we should

   “look at [psychologists’] explanations, but we should also keep in mind how well they cohere with the best account of what many psychologists really mean when they say “[central executive control]”.

   That certainly is what I take myself to be doing.[xvi] I hope that impressions to the contrary were mainly due to the all-too condensed presentation of my views on the central executive system in the paper.

    

   To conclude, I do believe that we have a fairly good grip on certain aspects of the central executive system. I do believe that there is a conception of that system that probably most psychologists working in the field could agree on. Progress in the field, as in many parts of psychology, has been impressive. It is time for philosophers to engage with this research and to use it to address philosophical issues. One way of doing so, I continue to believe, is explicating guidance by appeal to this research.

    

   Reply to Watzl: Primitive action, reflex action, and appeal to empirical research

   There is much that Sebastian and I agree upon, I think. I agree, for example, that attention shifts are mental acts; I agree that much of the discussion could have been framed in terms of the nature of guidance; I also agree with Sebastian’s observation about attention’s not being a necessary element in agency. Some of his questions are on topics I did not have opportunity to talk about in the paper and that I would like to address more fully in future work. Let me try to answer some questions that come up in Sebastian’s comments.

    

   Is action theory talking about a different, more sophisticated phenomenon, than what I am interested in? I do not think so. I believe that the core of our concept agency is that an individual does something, is the source of a process. Less fundamentally, an acting individual aims at something. Both phenomena can be encountered across the animal kingdom. An account of agency cannot ignore this fact.

   I partly agree with the stance Sebastian proposes. We can get the most out of many extant accounts if we use them to understand more sophisticated types of action. The accounts have difficulties there, too. Some of them I mention in the paper. Even humans plausibly exercise more primitive agential capacities than some of the accounts acknowledge. Further, in order to fully explain, for instance, perception-guided action, we must address the attributability-constraint. Extant accounts do not do that. So they do not even fully explain the phenomena that are clearly supposed to be in their purview.

   But I also think that the focus on sophisticated human action betrays a more general tendency that we should counteract. A too exclusive focus on humans leads to false theorizing even about some of our own abilities. Such a focus skews our view of our own place in the natural world. Understanding our capacities as continuous with those of other animals may not merely help us avoid making such mistakes. It may open new avenues for the scientific and philosophical study of these capacities. Primitive action is significant or of interest because to fully understand agency, I think, we need to understand it in all its breadth.

    

   What is the role of the central executive system in routine and reflex action? This is an excellent question, and my answer to it may be somewhat disappointing. In much routine action, I believe, the central executive system plays a background-organizing role. Take the case where you walk up and down your room while thinking about your paper. The central executive system activates or triggers your walking as much as your thinking, coordinates the two activities while you’re executing them, and allocates resources to them.[xvii]

   How about reflex actions? Think of the boxer’s weaving a jab, or your pulling away your hand from the stove; distinguish these cases from mere reflexes, such as a knee jerk. It seems plausible that the former episodes are acts. Do they involve executive control? I can see several ways of accommodating such actions in the present framework:

   1. They may involve minimal executive control, similar to routine actions.
   2. They may be directly available to central executive control (as in Frankfurt’s driving-example) while not concurrently requiring exercise of such control for their execution.
   3. Acquiring the ability to react in any of these ways may require executive control, while subsequent performance can occur quasi-reflexively.
   4. They may constitute a class of non-guided reflex acts for which my picture cannot account.

   Naturally I have a preference for options 1-3, since each of these options would allow me to explicate (mammals’) agency in terms of a process’s relation to executive control. Option 4 would make such an explication more difficult. On this option, I might attempt to argue that some reflex-‘acts’ are not really acts. For others it may be plausible to argue that they are not central to our notion of action. I then need to give an explanation as to why these are acts nevertheless. In the end, a full answer to Sebastian’s question is, at least in part, an empirical matter. I am currently working toward such an answer. For the time being I suspend judgment as to which option is the correct one.

    

   What does the central executive system explain? I say more about the status of research on central executive control in my reply to Mark. If our aim is the psychologist’s, namely, to explain, for instance, how an individual can solve the Tower of Hanoi test, then appeals to central executive control may not explain much. We do not understand well the mathematical and other skills involved in solving the test. We do not understand well how the central executive system helps us to successfully complete the task. We are far from having full computational or neural models, or a full mechanistic explanation of the processes underlying these tasks. It is in this sense, I think, that Baddeley considers the central executive system a “placeholder.”

   My claims about explanation are much more modest. There is empirical reason to think that there is a central executive system that helps coordinate and integrate individual’s capacities. There is empirical reason to think that the system does this executing a set of executive functions. By making connecting executive control with guidance we can both sharpen the notion and make present and future psychological research bear on our understanding of the notion.

   Sebastian is right to point out that not all executive functions are involved in every active attention shift. But some of them are (or so I claim). For instance, a spontaneous active shift may involve switching from one act to the next. The characterization of the system as an integrator and coordinator is rather of relevance in explaining why this involvement of the system explicates the individual’s guidance. In this sense, it is appeal to this system’s activity that helps us explicate guidance, not appeal to the individual’s activities.

    

   Can we establish my claims about guidance in an a priori manner? I like Sebastian’s idea and was initially very attracted to it myself. But I don’t believe that my specific proposal can be had quite that easily. In some sense, your metabolism is a structure that integrates and coordinates activity in different parts of yours. But your metabolism doesn’t yield the right kind of whole individual. Since I cannot think of an a priori way of spelling out what the appropriate structure must be, I attempt to identify it in an a posteriori manner. I’d be quite happy if I turned out to be wrong on this point.

    

   Notes

   [i]           Yantis, S. & Jonides, J. 1990. “Abrupt Visual Onsets and Selective Attention: Voluntary versus Automatic Allocation.” Journal of Experimental Psychology: Human Perception and Performance 16: 121-134; Theeuwes, J. 1991. “Exogenous and Endogenous Control of Attention: The Effect of Visual Onsets and Offsets.” Perception & Psychophysics 49(1): 83-90; Yantis, S. 1993. “Stimulus-Driven Attentional Capture and Attentional Control Settings.” Journal of Experimental Psychology: Human Perception and Performance 19(3): 676-681; Here are some more recent sources: Wright, R. & Ward, L. 2008. Orienting of Attention. Oxford: OUP; Carrasco, M. 2011. “Visual Attention: The Past 25 Years.” Vision Research (51): 1484-1525; Ruz, M. & Lupianez, J. 2002. “A Review of Attentional Capture: On its Automaticity and Sensitivity to Endogenous Control.” Psicologica 23: 283-309; Forster, S. & Lavie, N. 2008. “Attentional capture by entirely irrelevant distractors.” Visual Cognition 16; Kawahara, J., Yanase, K. & Kitazaki, M. 2012. “Attentional capture by the onset and offset of motion signals outside the spatal focus of attention.” Journal of Vision 12: 1-13

   [ii]           Cf. p. 208 of Corbetta, M. & Shulman, G. 2002. “Control of Goal-Directed and Stimulus-Driven Attention in the Brain.” Nature Reviews Neuroscience 3: 201-215

   [iii]          The best models concern free viewing or tasks in normal scenes that do not contain such extreme stimuli as the ones mentioned earlier. Cf. Geisler, W. & Cormack, L. 2011. “Models of Overt Attention.” In S. Liversedge, I. Gilchrist & S. Everling (eds.), The Oxford Handbook of Eye Movements (pp. 439 – 454), New York: OUP; cf. also Itti, L. & Koch, C. 2000. “A Saliency-Based Search Mechanism for Overt and Covert Shifts of Visual Attention.” Vision Research 40: 1489-1506; Itti, L. & Borij, A. 2014. “Computational Models: Bottom-Up and Top-Down Aspects.” In A. Nobre & S. Kastner (eds.), The Oxford Handbook of Attention (pp. 1122-1158), Oxford: OUP.

   The Reynolds model, as I have understood it, is a model of the attentional modulation of the activity of individual neurons. It is not a model of attention shifts. Why does Peter think that it bears on the issue at hand?

   [iv]          Hayhoe, M. & Ballard, D. 2005. “Eye Movements and Natural Behavior.” Trends in Cognitive Science 9: 188-194; Ballard, D. & Hayhoe, M. 2009. “Modeling the Role of Task in the Control of Gaze.” Visual Cognition 17: 118501204; Land, M. 2009. “Vision, Eye Movements, and Natural Behavior.” Visual Neuroscience 26:51-62

   [v]           Indeed, there was a fairly extensive debate about how exactly to characterize attention ‘capture’. Some psychologists proposed to characterize it in terms of search-slopes, others in terms of automaticity. Neither characterization seems to clearly rule out that individuals might guide the relevant processes. I side-step this debate in my paper. For a fuller discussion of capture and the issues in this and the last paragraph, cf. Buehler, D. 2014. Psychological Agency – Guidance of Visual Attention, UCLA Dissertation, Chapter 3.

   [vi]          Note that even if it turned out that all attention shifts are influenced by endogenous factors, the proposal I make as to which endogenous factor correlates with guidance might still turn out to be true.

   [vii]          Najemnik, J. & Geisler, W. 2009. “A Simple Summation Rule for Optimal Fixation Selection in Visual Search.” Vision Research 49: 1286-1294; Zelinsky, G. 2008. “A Theory of Eye Movements during Target Acquisition.” Psychological Review 115(4): 787-835; Pomplun, M., Reingold, E. & Shen, J. 2003. “Area Activation: A Computational Model of Saccadic Selectivity in Visual Search.” Cognitive Science 27: 299-312; Eckstein, M. 2011. “Visual Search: A Retrospective.” Journal of Vision 11(4): 1-36

   [viii]         A fuller discussion of recent models of visual search and their neural underpinnings can be found in chapters 3 & 4 of my 2014 Psychological Agency – Guidance of Visual Attention, UCLA Dissertation.

   [ix]          Cf. the literature in Fn. 7

   [x]           Mark’s worry partly overlaps with one of Sebastian’s, I believe. For a fuller treatment, cf. chapter 2 of my Buehler, D. 2014. Psychological Agency – Guidance of Visual Attention, UCLA Dissertation

   [xi]          I believe that philosophers have tended to ignore or downplay research on the system due to a set of influential criticisms of this research by Allan Allport and Daniel Dennett in the 90s. (cf. Allport, A. 1993. “Attention and Control: Have We Been Asking the Wrong Questions? A Critical Review of Twenty-Five Years.” In D. E. Meyers & S. M. Kornblum (eds.), Attention and Performance XIV: Synergies in Experimental Psychology, Artificial Intelligence, and Cognitive Neuroscience (pp. 183-216). Cambridge, MA: MIT Press; Dennett, D. 1994. Consciousness Explained. Boston: Little, Brown & Co; Monsell, S. & Driver, J. 2000. “Banishing the Control Homunculus.” In S. Monsell & J. Driver (eds.), Control of Cognitive Processes: Attention and Performance XVIII (pp. 3-32), Cambridge, MA: MIT Press). Current rejections of this research are typically not based on a serious engagement with recent results in the field.

   [xii]          Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., Howerter, A., & Wager, T. D. (2000). “The unity and diversity of executive functions and their contributions to complex “frontal lobe” tasks: A latent variable analysis.” Cognitive Psychology, 41: 49–100

   [xiii]         Friedman, N. & Miyake, A. 2004. “The Relations among Inhibition and Interference Control Functions: A Latent-Variable Analysis.” Journal of Experimental Psychology: General 133(1): 101- 135

   [xiv]         Gazzaley, A. & D’Esposito, M. 2006. “Unifying the Prefrontal Cortex: Executive Control, Neural Networks, and Top-Down Modulation.” In B. Miller & J. Cummings (eds.), The Human Frontal Lobes. Functions and Disorders. (pp. 187-207) New York: Guilford Press; Munakata, Y., Herd, S., Chatham, C., Depue, B., Banich, M. & O’Reilly, R. 2011. “A Unified Framework for Inhibitory Control.” Trends in Cognitive Sciences 15(10): 453-459

   [xv]          Anderson, V., Jacobs, R. & Anderson, P. (eds.) 2008. Executive Functions and the Frontal Lobes. A Lifespan Perspective. New York: Taylor & Francis; Conway, A., Jarrold, C., Kane, M., Miyake, A., & Towse, J. (eds.) 2007. Variation in Working Memory. Oxford: OUP; Friedman, N., Miyake, A., Young, S., DeFried., J. Corley., R. & Hewitt, J. 2008. “Individual Differences in Executive Functions are almost entirely Genetic in Origin.” Journal of Experimental Psychology: General 137(2): 201-225; Munakata, Y., Snyder, H. & Chatham, C. 2012. “Developing Cognitive Control: Three Key Transitions.” Current Directions in the Psychological Sciences 21(2): 71-77; Kurzban, R., Duckworth, A., Kable, J. & Myers, J. 2013. “An Opportunity Cost Model of Subjective Effort and Task Performance.” Behavioral and Brain Sciences 36(6): 661-726; Zaitchik, D., Iqbal, Y. & Carey, S. 2014. “The Effect of Executive Function on Biological Reasoning in Young Children: An Individual Differences Study.” Child Development 85(1): 160-175

   [xvi]         Cf. fn. 11.

   [xvii]         Baddeley and others thought of the central executive system activity as necessarily involving attention, even consciousness. This is not the notion I am interested in. I am more interested in the central executive system qua coordinator and integrator that also has limited resources available for its work.

   Loading...

   1. Thanks for your reply, Denis! I wanted to follow up a bit. You say that the claim that appeals to the central executive are not explanatory is an outdated one. But it seems to me that on Baddeley’s way of thinking, it’s just a central part of the concept of the central executive that appeals to it won’t be explanatory. So it’s hard to see what sort of, say, empirical discovery would outdate the claim. Instead, something rather central about the concept itself would have to change.

      About breaking down the central executive into components and then explaining those components: this does seem like progress. But – as I understand Baddeley’s way of thinking – the thought is that once psychologists have identified and explained all those components, then they should stop making use of the concept of the central executive, and instead use the concepts of those better understood component parts.

      Is your thought that he is wrong about this, and that there is some sort of way these components are unified that would justify continued use of the concept of the central executive?

      Loading...

      1. Maybe we think differently about how concepts such as that of the central executive system work. You seem to want to appeal to something like ‘Baddeley’s definition of the concept’ when you say that empirical discovery couldn’t show that the concept is explanatory after all. Do I get that right? (I’d be interested to hear more about your views on this issue!) I am not sure that this is how concepts in psychological explanations work. Here is how I tend to think about the relevant concept. Baddeley introduced a notion into the debate (actually, it may be due to Broadbent, in this case). In doing so, he ‘got on to’ a psychological system – a set of executive functions for the control of cognitive processes. Both the referent of the notion and the referent’s characteristics have been precisified in subsequent research.

         As I pointed out, there has already been progress identifying sub-components of the system. I mentioned the executive functions of switching, maintenance, and inhibition. Doesn’t that mean that, already, we can make do with talk about the central executive system and instead confine ourselves to talk about these three functions? In some cases, we may explain all that we want to explain by doing so. In other cases, grouping the components together may be illuminating. There is currently no reason to think that executive functions, as we now think of them, will not exhibit sufficient unity for us to group them together and refer to them as belonging to the same kind. But is there any positive reason for treating them in this way? One reason is that they all function to control the execution of cognitive processes. Another reason is that there is behavioral evidence for unity among the executive functions. Apparently, the execution of different executive functions taps into one and the same underlying resource. Finally, there is reason to think that one prominent neural network – involving prefrontal cortex – implements executive functions in mammals.

         This is not to say that we might not turn out to be wrong about this. Elements of our theories of the executive system may well turn out to be mistaken. Maybe future science gives us reason to think that there really is no such thing as a set of executive functions. Maybe it will turn out that the apparent unity that we seem to see now was an illusion. My point is that there currently seems to be no reason to be worried about this possibility.

         Loading...

         1. Thanks, that’s helpful! To continue a bit – you say:

            “Here is how I tend to think about the relevant concept. Baddeley introduced a notion into the debate (actually, it may be due to Broadbent, in this case). In doing so, he ‘got on to’ a psychological system … ” .

            What I find surprising about this is that in your view, with Baddeley’s use of ‘the central executive’, he got on to a single psychological system despite his explicit intention not to do so. He seemed to want to use ‘the central executive’ to refer to a gerrymandered set of things. I don’t think that your interpretation of what’s happened is necessarily incorrect. But I do think that we should want an account of how and why the referential intentions of Baddeley and other psychologists have failed in this surprising way.

            I know that you talk more about these issues in your second chapter, so I will take a look – it’s very interesting!

            Loading...

            1. Hey Mark, what do you base your diagnosis of Baddeley’s referential intentions on? I am not so sure, if you look at earlier work, that he had any intention NOT to discover a unified system or set of functions. He just seems to be ok with that possibly turning out to be so. Also, I appeal to research by Miyake and others. So if anything, their referential intentions should matter.

               But again, the picture of scientific concepts that you at least seem to suggest strikes me as not particularly compelling. I do think that what such concepts mean or are about is not determined by the intentions of the guy who first used the word that expresses the concept, but by the kind that the concept gets on to (and to some extent, its use in the scientific community). What do you think about this?

               Loading...

5. This is a comment on Peter Fazekas’ comment. Peter, you imply that no attention shift is a form of guided behavior in the relevant sense, or that – as you put it – “shift of attention is always captured by the stimulus with the highest salience.” The evidence you cite simply does not establish this, or even make it plausible. What the evidence suggests is that attention shifts are controlled by a priority map that always gets input both from bottom up as well as from top down. But that simply doesn’t undermine that some attention shifts are guided by the subject or suggest that attention is always captured.

   (What it probably does suggest is that the line between behavior that is individually guided and behavior that isn’t is either quite blurry or maybe even a matter of degree – of which I favor the latter).

   Consider a standard display: two identical Gabor patches, one left, the other right (nothing else on the computer screen). Suppose that there is a pointer that runs around a clock (one turn per minute) – like in the Libet experiments – though you don’t really need the pointer. The subject freely decides when to shift attention either left or right. There is no evidence that subject’s cannot do this (a lot of evidence that they can). Of course, like in all cases there are lots of factors (prior reward and experience among them) that will influence which way subject’s will go (that’s what the evidence shows, and – indeed – that’s what determinism implies), but that does not show that the shift wasn’t guided by the subject, or that this is a case of (passive) attention capture. I can’t see how one could make a claim like Peter’s about a case like this, without also claiming that the subject in a Libet experiment isn’t actively pressing the button (i.e. that pressing the button isn’t an actively guided form of behavior). The interplay between bottom up and top down factors (both of which are not a unified class), will be present in the control of bodily behavior just like in the case of attention. Indeed, just take saccading as your bodily behavior and we know all the relevant evidence. I take it as a reductio of the Peter’s claim that no attention shift is a form of guided behavior that on his view no saccade is a form of guided behavior either. So, Denis main example still stands.

   Loading...

6. Thanks for this great response, Denis. Very helpful. With respect to the routine actions: I think we should accept that as actions become more routinized they require less and less central executive control (also no background organizing role). Think of running on a path you’ve run a thousand times before, or playing a piece of music you’ve played a thousand times before. While doing those things your mind might be free to do lots of other things. For those routine actions (like for what you call reflex actions) we need one of your other three options (2-4). I think 3 + 4 are giving up too much, so I’d like to know a bit more about 2. I worry that “easily available for executive control” isn’t so well defined. Think of the running case. I am not sure it is easy to apply executive control to running. Overlearned activities are often hard to change or control. So, I’d like to hear a bit more hear.

   Loading...

   1. As I said, I am a little less sure which of 1-3 will turn out to be true. You seem to be more committal here. Is that right? What is the commitment based on? I am wary of equating central executive control with attention or consciousness. And I am wary of trusting my intuitions about cases such as your running example. It seems to require empirical research to decide whether the central executive does not after all play some background organizing role, even in such episodes. Or do you think I am overlooking something, here? Really, what I am curious to learn more about is, what kind of evidence do you expect will decide the matter?

      I agree that the remarks about ‘easy availability’ are – as you put it so diplomatically – less than well-defined. The way I was thinking was very much along the lines you suggest in your first set of comments. There are old models (such as the Shallice 1988 model) of an executive system that is not involved in routine action, but will get involved as soon as a routine action encounters an obstacle or unexpected circumstances. Presumably, if there were some truth to these models, we should expect fairly close neural connections between networks responsible for routine action and executive control networks. Presumably, ‘availability’ would entail fast access of executive control to routine action, when needed. Appeal to these models does, of course, not do much to sharpen the idea of availability. But they may provide a start. Does that help a bit? Any idea how one might do better?

      Loading...

   2. Thanks to Denis for an extremely interesting paper, and to the commenters for their in-depth discussion.

      The topic of ‘routinization’, or automatization, of actions touches on issues of skill-acquisition, and this is an interesting point of contact between this discussion and others in the session.

      I’d like to suggest that something like Denis’ options 1 or 2 holds promise. It seems that routinized actions, however overpracticed, still must be available for some executive guidance, insofar as the well-trained guide is able to manipulate higher-order properties of the performance, or to avert novel and unexpected performance errors.

      In Sebastian’s example of playing a well-known piece of music, the guide retains the ability to ‘do things’ to the performance (e.g. play the piece faster or slower, alter its dynamics by highlighting different notes, etc.), and the more routinized the piece, the more available these modifications should be. And the guide is also able to recognize when something is going wrong: in Sebastian’s other example, the running path may have some new puddles from last night’s rain, or garbage that someone left there, or may be used by other people, and well-trained agents should be able to avoid these obstacles.

      For both manipulating higher-order properties and averting obstacles to performance, some sort of executive monitoring seems to be necessary. This monitoring may work by (1) executive control retaining some ongoing minimal involvement in performance, or by (2) the performance being directly available to executive control. But at any rate, even in the most routinized cases, the agent seems to retain some level of improvisational and corrective ability, which seems rather hard to explain without appeal to some executive function.

      (Now, I don’t know whether we could say the same thing about what Denis calls “reflex actions”. Perhaps the strategy there should be to deny their status as ‘actions’, and just call them reactions… Pulling your hand away from a hot stove seems explainable via merely exogenous mechanisms.)

      Loading...

      1. Thanks Pablo for your remark! Do you know of research that explicitly addresses executive involvement in these cases? I do not think of executive functions as involving the individual’s reflection. Do you?

         Loading...

         1. Hi Denis!

            First, I just want to reiterate that it’s been a delight reading your paper and following the discussion.

            On the latter question: I defined ‘reflection’ as any process involving the executive functions, so… yes, but by definition. (In this I’m mainly following J. St. B. Evans (2010), but this session has made me question whether it’s worth it, just because it generates a lot of confusion.)

            Concerning empirical research explicitly addressing executive involvement in these cases, the first thing that comes to mind is the evidence for parsing that Ellen discusses in her paper.

            On a more conceptual point, Ellen just made the very interesting comment (in the thread on her paper) that, even if online executive control isn’t involved in a given skilled performance (say, because it’s quite familiar), the motor representations involved do retain a structure that was developed in previous instances of executive control. I found this to be an interesting idea, that may help illuminate how executive guidance (as opposed to control) is involved in habitual actions: even if not actively involved, executive control can get involved given that the motor representation has this intelligent structure, that parses it into multiple bits.

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

            Loading...

7. Thanks for the response Denis, and thanks for the further comment Sebastian.
   I am afraid my point has been misinterpreted, which, of course, is totally my mistake. So let me clarify.

   First, I have never intended to imply that “no attention shift is a form of guided behavior” (quote from Sebastian). I admit, the sentence “shift of attention is always captured by the stimulus with the highest salience” that Sebastian quotes is easy to understand in this way, but it was intended to be in line with the preceding paragraph in the original comment, in which I describe what we know about the mechanism behind shifts of attention. We all agree that exogenous factors, executive control, and other endogenous factors together determine salience values, after which attention is shifted to the stimulus whose representation has the highest salience on the salience map (this is the sense in which ‘salience’ in the sentence quoted by Sebastian was intended to be read, not in the sense of physical salience). In accordance with this, the unfortunate use of ‘capture’ was not a reference to bottom-up captured attention, but to the fact that once the salience map is computed, attention will be shifted to the stimulus with the highest salience value (on the map) automatically, regardless whether it is a stimulus targeted by the executive system.

   Second, I definitely do not need the strong claims that Denis and Sebastian read into my comment for my argument to work. My concern is the general correlation claim Denis makes: “central executive control correlates with primates’ guidance of attention shifts”. Denis’ own example is visual search. I argue that typically (or as I put it in the original comment “under realistic conditions”) in visual search, although the central executive system does play the role Denis attributes to it, shifts of attention do not seem to be guided. But if there are cases in which central executive control is in operation, but are nevertheless not instances of guided behavior, then it is not true that central executive control correlates with guidance, and it is not clear in what sense guidance could be explicated in terms of central executive control.

   So what is my problem with visual search? On the one hand, Denis, Sebastian and I are in agreement on the role of the central executive system. When attention lands on a non-target, the executive system initiates a reallocation, and modulates the computations that output the values of the salience map. So exactly the right kind of central executive control, which is supposed to help us understand guidance, is in play here. However, on the other hand, it only plays a role in computing salience values. The shifts of attention themselves are driven by the salience map itself: attention will be shifted to the stimulus with the highest salience value no matter which stimuli were targeted during the task. Denis thinks that it is enough to think of shifts of attention as being driven toward the goal. I disagree. Consider the set-size effect, i.e. the phenomenon that in visual search response time increases as the number of distractors (non-targets) increases. With a lot of distractors present attention lands on non-targets several times before it finally lands on the target. Is this series of non-target hits such that the non-targets consecutively attended can be said to be closer and closer to the target according to any metrics based on any feature space? I am not aware of any results supporting this claim. Or think about playing ‘Where is Waldo?’. The series of misses before spotting Waldo doesn’t feel like gradual progress, or being driven toward finding him. It feels more like trial and error. Shifts of attention in such (typical) cases of visual search do not seem to be driven toward the target.

   Lacking any clear definition of guidance Denis asks us to rely on our intuitions. I do not share Denis’ intuition that a tour guide offering such a trip is someone who I should follow. And if instead of our intuitions we try to rely on Frankfurt’s original characterization of guidance, the situation, as I have argued in the original comment, is no better. I conclude that shifts of attention in important cases of visual search (where executive control is nevertheless in play) are not guided.

   A final word on Sebastian’s reductio. As I’ve tried to emphasize above, the crucial thing is not that there is an interplay between bottom-up factors and executive control, but rather how the interaction works and what role the contribution of executive control plays in the actual mechanism. According to my interpretation, in typical cases of visual search the contribution of goal representations in directing shifts of attention is not enough for such shifts to be labeled guided behavior.

   Loading...

   1. Hi,

      First off, Denis: I find your presentation (and your research more generally) extremely interesting. I have a huge number of questions for you, and obviously I have a lot to learn from reading your work! I’m going to try reading your dissertation this week, as it’s relevant to what I’m working on now.

      I just wanted to ask a question about the choice of visual search as an example of individual-guided behaviour. I’ve been reading Peter’s comments and I’m sympathetic to some of the points he’s been raising about the automaticity of (most?) attention shifts in realistic conditions, and about the way some researchers think about the role of the salience map or priority map in determining how they occur. (The terminology ‘salience map’ and ‘priority map’, by the way, isn’t well-regimented as far as I know. I think some researchers seem to use these terms interchangeably, while others seem to use them as theoretical competitors, so that the priority map is meant to replace the older notion of a salience map by giving a more central role to broadly ‘top-down’ factors in determining the state of the map. I’m curious if anyone disagrees and would like to explain this for me, since I’m keen to know how these terms get used in the empirical literature).

      So, here’s my question: if Denis is looking for an intuitively compelling example of individual-guided attention, then I wonder if visual search tasks are really the best candidate, and I wonder if the choice of example could be giving Peter the upper hand in this conversation. Just intuitively, the problem with visual search (it seems to me) is that the ‘guide’ doesn’t know where the target is. Background knowledge and the representation of the goal can influence the efficiency of the search in various ways, but it seems to me hardly analogous to the mountain guide, who (let’s suppose) has a lot of knowledge about where the goal is in relation to him, the different routes that are available to him, etc. Just intuitively, there will be a lot more trial and error or stabbing in the dark involved in a Where’s Waldo-type search than in many other sorts of task. And there are other kinds of tasks that we could consider: e.g., ones investigated in ‘natural vision’ studies in which subjects perform well-practiced tasks in familiar, well-understood environments (e.g., making a sandwich, playing a ball game, etc.) or, to use an example Wayne Wu likes to discuss, looking at a painting with the aim of answering questions about it (Yarbus). The results of these eye-tracking studies are always quite striking: the saccades (and hence attention shifts) are manifestly goal-directed. (And some researchers have recently been arguing that those results aren’t well-explained by ‘salience-based’ accounts of attention). So, I’m wondering: if we focus on a different kind of visual task — one in which the central executive is intuitively in a better position to act as a guide, given background knowledge about the scene and where the relevant sources of information are likely to be found — will this help Denis in this debate with Peter? Unless I’m underestimating the power of the central executive (which I might be!), it just seems unfair to expect a ‘guide’ to direct attention toward the target in such a situation, given that the guide doesn’t know where it is. This doesn’t preclude the use of other intelligent strategies for getting there more quickly than one would otherwise, but we can’t expect the guide to do what’s being asked of it. On the other hand, I would think there are lots of other tasks in which the central executive is well-positioned to shift attention toward task-relevant information and not just to the most salient location. And maybe in those other situations the attention shifts do satisfy Peter’s demands on what it would take for attention to be guided.

      Loading...

      1. Hi Aaron, thanks for your interest! Let’s def be in touch, if you want. I’d like to hear about your work, too.

         First of all, yes, I use ‘priority map’ deliberately, because I think it better captures the idea that many different factors – other than the traditional physical saliency – can influence what locations have priority for shifting attention to them.

         Now to your question. I still think that the case of visual search is a good case. It seems (a) clear that we guide visual search, even though (b) we cannot spell out guidance in terms of ‘knowing the destination.’ Finally (c) (visual) search is of absolutely central importance to so many of our activities.
         We are clearly not randomly moving attention around when we engage in visual search. Behavioral data and computational models (have a look at the Geisler references, if interested) show that our attentional system searches in highly efficient and successful ways. True, sometimes it may feel as though we’re just randomly looking around in desperate search for our keys, but actually many factors contribute to making our search more efficient.
         People often tend to think of guidance toward a goal in terms of knowing the destination. That is a natural way of thinking, but it is also too simplistic, I believe. One the one hand, of course, there is an element of that also in visual search. I do know what my keys look like, I look for similar things, and I do so until I’ve found them (I also may apply strategies for finding them, but bracket that). On the other hand, some of the most important activities that we set out to accomplish – foraging, remembering, problem-solving – are such that we cannot have full knowledge of their goals. Otherwise they would be pointless. But they do neither feel random, nor are they, as a matter of scientific fact, random.
         Our embarking on any of these activities activates and shapes subsequent search processes in ways that, by and large, allow us to be fairly successful at reaching our goals. And it is easy to appreciate how important these activities are for understanding what we are up to on a day to day basis.

         Loading...

         1. I see. Thank you, Peter and Denis. I jumped in without properly understanding the dialectic. I think I now see that my point was actually supporting Peter (in fact, possibly repeating him!), and that my observation doesn’t help Denis in this context. I’m not sure where my sympathies lie just yet!

            Stepping back, I’m curious to get clearer on the disagreement between you two. Peter’s claim (correct me if I’m wrong) is that executive control doesn’t correlate with the individual’s guidance, because, while executive control is indeed involved in controlling attention during visual search (by influencing the state of the salience/priority map), it doesn’t guide attention to its goal. Denis’s claim is that they do correlate, and that central executive does guide attention in visual search. Assuming that what Peter means by ‘salience map’ is what Denis means by ‘priority map’ and that you both agree about how executive control informs the state of that map, the disagreement between you two seems to be about how to understand guidance. In particular, I have a feeling this passage from Peter is revealing about how Peter thinks about guidance, and where I suspect Denis can jump off:

            “Consider the set-size effect, i.e. the phenomenon that in visual search response time increases as the number of distractors (non-targets) increases. With a lot of distractors present attention lands on non-targets several times before it finally lands on the target. Is this series of non-target hits such that the non-targets consecutively attended can be said to be closer and closer to the target according to any metrics based on any feature space? I am not aware of any results supporting this claim. Or think about playing ‘Where is Waldo?’. The series of misses before spotting Waldo doesn’t feel like gradual progress, or being driven toward finding him. It feels more like trial and error. Shifts of attention in such (typical) cases of visual search do not seem to be driven toward the target.”

            There are a few ideas here, but one of them is the assumption that guidance control entails “gradual progress” toward a goal. Given the implicit contrast with trial and error, I take it that gradual progress doesn’t simply consist in the elimination of possibilities (e.g., “Is this the target? No. Is this it? No. This? Nope …”), since the latter looks like trial and error (combined with some memory of which locations one has checked). Given that gradual progress doesn’t simply consist in the elimination of possibilities, it would be worth clarifying what this gradual progress requirement consists in and whether it’s essential for guidance. I have a feeling, from what Denis said in reply to me, that he would reject the requirement (cf. Denis’ denial that guidance requires knowing one’s destination. He wants guidance to accommodate more open-ended instances of problem solving).

            Loading...

   2. My idea was that visual search is a clear case in which individuals guide their attention shifts toward a goal – the goal of their search. Suppose that I am searching for a banana. How is my guidance implemented? Executive control implements my setting of the search goal (the banana) and thus controls computations of priority on the priority map (based on similarity of stimuli at locations in the scene to the desired banana). Attention will shift to locations based on priority values for different locations on the map. Similarity to the banana will shape the computations for where to shift next until the goal of the search is reached – until I’ve found the banana. There is no guarantee that attention will shift to the right location (the one with the banana) immediately, or without fail. Indeed, attention often gets it wrong. How wrong attention gets it (and how long it takes to find the banana) depends on the size of the sample, the number of distractors, noise in display and system, and many other factors. But computational models of visual search tell us that the process is very far from being trial and error (cf. the references in my last reply). It is in virtue of the way in which executive control shapes how priority is computed, that we can say that attention is driven toward the search goal.

      You say that you disagree with my proposal. I am not entirely sure I’ve understood why you disagree. Do you think that, intuitively, attention shifts during search are not guided? Do you think that any automatic process could not be guided? Do you think that any process that misses its goal (maybe by a lot) could not be guided? Do you think that any process involving exogenous factors could not be guided? None of these commitments strikes me as promising. Or were exactly do we disagree?

      Loading...

      1. Sorry, this last reply was for Peter …

         Loading...

      2. Hi Denis,

         We agree on how executive control operates in visual search, and how it is able to modulate the computations producing salience or priority values. I think we disagree whether this is enough to implement guidance of shifts of attention. I don’t think that it is. What such a modulation can achieve is that it ‘pushes’ the target representation higher up on the priority list. Let’s say that the top is position ‘1’, and executive control modulation can push target representation to position ‘n’ (whereas without such a modulation is would be lower, say in position ‘n+k’). Now even if it is true, I think that shifts of attention during search as it is reallocated to locations corresponding to positions from 1 to n, and finally settles on the target is guided neither in an intuitive, nor in a Frankfurt-based sense. Not because they are random — they are not, that’s exactly what the priority map ensures — but because, to use your criterion, they do not get ‘closer’ to the target (not just spatially, but according to any metric) as they shift from position n-i to n-i+1. Or put it in Frankfurt’s terms: as attention shifts in correspondence with moving through the priority list from position 1 to n “its course is not subject to adjustments which compensate for the effects of forces which would otherwise interfere with the course of the behavior” (Frankfurt, 1978, p.160).

         Loading...

         1. Hi Peter, I posted a comment above the same minute you posted this one, and with significant overlap. Sorry for the repetition!

            Loading...

         2. Hey Peter,
            I did not mention the criterion (getting closer to the target by some metric) that you attribute to me. It is not clear to me that it is a condition on guidance. It so happens that it can be easily met in the case of visual search, though. I’ll get back to that below.

            I’d be mostly curious to hear your answer to the questions I had for you in my last post: on what basis do YOU deny that search is guided? Or would you say that the above mentioned criterion does capture your view?

            Let me repeat some points from earlier:
1. The attentional system is set so as to be ‘looking for’ things similar to the banana (task-driven attentional routines like the ones that Wayne and Aaron mention often are, or involve visual searches; so if you’re not acknowledging guidance in visual search, you should not do so in many of those cases either).
2. The search performed on this basis is highly efficient and increases the probability of finding the target vis-a-vis trial-and-error searches, random searches, or merely saliency-based computations. Indeed, the probability of finding the target with the NEXT shift increases with each shift. We learn this from the most advanced models of visual search that I quote in my earlier replies. Here’s an obvious sense in which attention gets ‘closer’ to the target, if you like that criterion. (The relevant computations also suppress the influence of e.g. physically salient distractors on where attention shifts. So here’s also a sense in which “its course is … subject to adjustments which compensate for the effects of forces which would otherwise interfere with the course of the behavior,” as Frankfurt suggests it should be.)
3. There is no reason to think, certainly none that you have provided, that under these circumstances we would not speak of guidance - or do you seriously want to deny that someone could not guide the search-expedition for Pablo Escobar, unless they already knew where Pablo Escobar is hiding? (Anybody’s guess what show I’ve been watching last night.)


            So again, how are these considerations mistaken? What precisely are the alternative considerations in favor that visual search is never guided?

            Loading...

8. Hi Aaron,
   Thanks, great comment! It illuminates really nicely my point that the source of the problem is what role executive control can play in the actual mechanism of shifting attention.
   I certainly think that my worries apply only to the specific case of visual search, so I am open, and, in fact, sympathetic to the idea that there are other visual tasks (like Wayne’s example) where shifts of attention are indeed guided, and executive control (restricted to what it does in those cases) is in a better position to correlate with guidance.

   Loading...

9. Hi Denis,

   I’m sure you’re busy answering all of these other comments, but I just had a quick question of clarification about your view. And my apologies if you answer the question in your paper.

   What’s the relationship, on your view, between an agent’s intention to act in some way and their central executive system? A natural view, it seems to me, would be to link the two, so that the central executive helps implement or realize an agent’s current intention. Your objections to traditional action theories seem to suggest that you would not be sympathetic to such a proposal, since you think that intentions are too intellectually sophisticated to explain the agency of non-human animals. What do you say to someone who just equates intention with the executive functions (or some subset of those functions) carried out by the executive control system? That would, it seems, bring your empirically-based view into line with a fairly traditional causal theory of action, on which an agent’s guidance of action consists in the guidance of behaviour by an intention. It would be a consequence of the proposal that many non-human animals have intentions (given that they have a central executive system), but that seems sensible enough to me. Or do you think there are good reasons to deny that non-human animals possess intentions?

   Loading...

10. I’m also curious what you would say to someone who objects like this: central executive control guides both active and passive attention shifts. The difference between active and passive attention concerns which executive functions are present and which are absent. In active shifts, attention is guided by a goal representation which is maintained in working memory and kept on course by the inhibition of goal-irrelevant information (e.g., distractors). So, active attention exists where there is the presence of maintenance and inhibition and an absence of task-switching. Passive shifts, by contrast, are characterized by the opposite pattern of executive function: an absence of maintenance and inhibition and the presence of task-switching. That is, when your attention is captured, you switch tasks. Assuming that passive attention isn’t agent-guided, then the guidance of attention by executive control doesn’t correlate with agent-guidance.

    Perhaps one could substantiate the details of the proposal by drawing on Corbetta & Shulman’s (2002) distinction between the dorsal attention network and the ventral attention network. The idea would be that the ventral network specializes in task-switches in response to behaviourally significant stimuli, whereas the dorsal network specializes in keeping the agent on task. (Though maybe this way of filling out the empirical details is false).

    Loading...

    1. Hey Aaron,

       On your first question, I would say that (a) I am not sure what it is to equate an intention with an executive function and (b) that, to the extent I get the proposal, it seems in conflict with the empirical facts. Executive control can involve propositional intentions or more image-like representations (and possibly neither). That is one of the strengths of my proposal, as I see it, for it allows us to encompass more animal agency with our account.

       On your second question, I would say that the proposal gets the facts wrong. I like your description of what’s going on in active shifts. But there’s no task switching that controls passive attention shifts. I do not think that appeal to the Corbetta 2002 distinction between different networks bears on this issue (remember the research that I base my characterization of the executive system on).

       I say more on both topics in the paper.

       Loading...

       1. Thank you, Denis. As I mentioned, I’ll be taking a closer look at your work soon.

          Just to elaborate on the first idea (about the potential connection between executive control and the subject’s intention), my thought was roughly this. At its core, you might think an intention is a type of representational state that plays a certain set of roles in the subject’s mental economy and in behavioural control. Which roles? One answer might be: the roles that you’ve identified with the central executive. If a goal representation plays the executive roles (e.g., if it’s maintained in working memory, if it inhibits goal-irrelevant information from consuming central resources, if it integrates and coordinates one’s other states and competencies in a way that tends, over time, toward its own fulfillment), it satisfies the functional description of an intention. Intention just is an executive goal representation.

          That’s the line of thought I was exploring. But if you’re right that states other than intention can play the same executive functions, then obviously the hypothesis I’m exploring is a non-starter. I just wonder what would disqualify those states from being intentions.

          Loading...

Comments are closed.