The Structure of Cognition: Attentional Episodes in Mind and Brain. Duncan. Neuron 2013.

  1. Opens with “Cognition is organized in a structured series of attentional episodes, allowing complex problems to be addressed through solution of simopler subproblems.  A ‘multiple demand’ (MD) system of frontal and parietal cortex is active in many different kinds of tasks… I propose a core role for MD regions in assembly of the attentional episode.”
  2. Neural studies show rapid communication between MD regions
  3. Work on modelling has mostly focused on increasingly fragmented issues – this paper tries to discuss these disparate behaviors in terms of the working of one system
  4. “I consider… the construction of all complex cognition from a series of focused, momentarily assembled temporal fragments.”
  5. Starts with discussion of “innate releasing mechanism” (IRM) which is an instinctual response to some stimulus.  “In much of animal behavior, concatenation of such behavioral fragments, each controlled by its own IRM, produces complex, goal-directed sequences or programs of activity…”
  6. “In human cognition too, thought and behavior unfold in a complex, structured sequence, with many component fragments assembled to achieve short- and long-term goals… Unlike the IRM, fragments of human cognition must be momentarily constructed, shaped by the arbitrary requirements of current activity.”
  7. Mention symbolic constructs such as ACT and SOAR
  8. Importance of chunking and hierarchy in planning in AI
  9. “As selection is the defining characteristic of attention, cognition may be described as a series of attentional episodes, with each episode admitting into consideration only the contents of a momentary, focused subproblem.”
  10. “…global disorganization of behavior that can follow frontal lob damage, often with intact behavioral fragments but no complex, goal-directed structure.” – each individual behavior is executed effectively, but the overall sequence of behaviors is not assembled in any manner that allows for a useful outcome
  11. “I review evidence for a specific, distributed frontoparietal system whose role, I propose, is to construct the sequential attentional episodes of complex cognition.”
  12. A list of brain regions making up MD that takes up about a paragraph – not listing them here.  They are commonly implicated in a wide variety of activities from fMRI studies
  13. “…these data show tightly localized MD activity, varying in exact pattern from one person to another but with a highly consistent overall topography in frontal and parietal cortex.”
  14. “I suggest that the core function of the MD system is to control complex behavior in a series of attentional episodes.”
  15. In the area of cortex they examine, across many different tasks “large proportions of cells code the specific information required in a current epoch of behavior.”  In one task they found more than half the neurons they monitored discriminated targets in a visual task from nontargets
  16. The fact that so many are predictive in so many types of tasks means there is a high level of flexibility
    1. “Correspondingly, information coding in prefrontal cells changes with momentary task relevance…. In working-memory tasks, cells show sustained activity linked to stimulus identity when this information is required within the trial but then switch to coding of location when identity is no longer relevant…”
  17. “A third critical property is flexible transition from one step of a mental program to the next.”  Some neural activity is specific to a particular target or goal, and some activates in the same pattern when a goal is presented regardless of what the actual goal is
  18. Coding for the same stimulus may also change over a trial, or over more extended periods of time <I suppose the implication is that this can happen independent of the other types of changes that can occur such as change in objective>
  19. <Not sure why this is the case but> “When cells adapt to code the specific information that is required in current behavior, the result should indeed be a pattern of widespread activity, irrespective of particular task content.”
  20. Sometimes activity patterns to generalize from one context to another, showing some type of common-coding <disagreeing with points made above, but of course the brain never cooperates cleanly with the way we would like it to>
  21. Strong MD activity occurs when transitions between tasks or events occur “a recent study shows extensive activity as each subgoal is completed, with increasing activity for subgoals progressively higher in the hierarchy (Farooqui et al. 2012). MD activity increases when materials for short-term memory are formed into higher-level chunks, perhaps reflecting activity as parts of a new chunk are bound into one whole.”
  22. Argues that the timescales of communication between regions in MD is probably too fast to pick up on fMRI, so coactivations are really the only way to study relationships with current technology, tiny changes in latency may also be task dependent
  23. “I suggest that MD activity binds together the components of an attentional episode, constructing the specific conjunction of processing events that current behavior requires.  Thus brian-wide activity is configured for solution to the current behavioral problem.”
    1. A model would need to keep track of <at least> the context, stimulus, and response
    2. “The system [model] has the most power when each randomly connected unit is bound to approximately half of the dedicated units”
    3. That model was implemented and has a number of properties that are similar to what actually happens
  24. Implication of frontal lobes in abstraction and attention, as both are often felt to be related
  25. If MD really does control so many different forms of thought, it would make sense that people who do well on one type of intelligence test will also do well on other classes of intelligence tests, which is what we see.
    1. Mentions Spearman’s “g” factor as a measure of this
  26. “The best measures of g are generally tests of novel problem solving, such as series completions, analogies, and matrices (…) commonly termed test of ‘fluid intelligence.’ ”
    1. Fluid intelligence is in contrast with crystallized intelligence which is based on facts that have already been learned
  27. Evidence from lesions shows that damage to losses in fluid intelligence are most closely related to the MD system
  28. <getting tired of reading, notes may suffer>
  29. Introduces idea of goal neglect, where “… the person understands and can describe task requirements, but these requirements seem not to control what is actually done… For example, the patient may be asked to squeeze the hand when a light is seen; when the light is switched on, the patient says ‘I must squeeze!’ yet in fact does nothing.”
    1. This is very common in individuals with frontal lobe damage
    2. Subjects describe it as themselves neglecting the task
  30. The critical factor in whether goal neglect occurs is not in the task itself, it is wrt initial task instructions – that is, if you tell them what to do piece by piece, they can accomplish each in isolation, but if they are told to do all the tasks upfront, they will fail – calls this a “complexity limit”
  31. When fluid intelligence is limited, it is difficult to perform well in tasks that are complex <author goes on about this, but that seems obvious?>
  32. “In cases of gross neglect of some task rule, we found that, most commonly, some fixed pattern of behavior replaced the correct rule.  This fixed but incorrect rule might take several forms: neglect of a critical task even, a rule reversal, a rule imported from another task, etc.”  Moreso, the rate of usage of the correct rule declined as trials progressed.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: