Thursday, October 05, 2006

Attention/ Memory/Learning: double dissociation between ACC and PFC

I recently came across two studies both of which were pointing towards a double dissociation between ACC and PFC, in the realm of Working Memory Attentional processes in one case and the learning mechanisms (or acquisition and performance of a cognitive skill) in the other case.

In the first study by Kane and Engle, a stroop interference task was used to find the different attentional factors at work that determine the successful execution of the task. Using some clever experiments, it was demonstrated that two selective attentional mechanisms were involved- one that was related to goal maintenance and was active pre-stimulus presentation and the other that was active post-stimulus presentation and was related to inhibition of inappropriate bottom-up responses (the automatic response as per the linguistic color-denoting word in incongruent condition instead of as per the actual color of the word that was demanded by the task) and selection of relevant response from the competing responses.

As per the abstract of the study:

Individual differences in working-memory (WM) capacity predicted performance on the Stroop task in 5 experiments, indicating the importance of executive control and goal maintenance to selective attention. When the Stroop task encouraged goal neglect by including large numbers of congruent trials (RED presented in red), low WM individuals committed more errors than did high WM individuals on the rare incongruent trials (BLUE in red) that required maintaining access to the "ignore-the-word" goal for accurate responding. In contrast, in tasks with no or few congruent trials, or in high-congruency tasks that followed low-congruency tasks, WM predicted response-time interference. WM was related to latency, not accuracy, in contexts that reinforced the task goal and so minimized the difficulty of actively maintaining it. The data and a literature review suggest that Stroop interference is jointly determined by 2 mechanisms, goal maintenance and competition resolution, and that the dominance of each depends on WM capacity, as well as the task set induced by current and previous contexts.

As per this line of reasoning, errors in the stroop task are thought to result from failure to actively maintain a goal in mind and may thus be related to memory retrieval per se. On the other hand, reaction time slowing is thought to result from a post-stimulus attentional process - a failure to quickly bias competition towards the correct representation rather than the incorrect representation and might also be perceived as an attentional control mechanism- whereby attention is not diverted to irrelevant stimuli that are not consistent with the goal in WM.

Developing Intelligence presents some additional observations to bolster the argument:

  • Across all subjects, the amount of RT facilitation (i.e., how much faster congruent trials are than neutral trials) correlates with error interference (i.e., how much more accurate neutral trials are than incongruent trials), suggesting that goal maintenance failure is behind both of these phenomena. In contrast, there is no correlation between the RT facilitation effect and RT interference, as would be expected if goal maintenance failure actually gives rise to all of these measures, nor is there a correlation between error & latency interference. The implication being that errors (and the related RT facilitation) are due to one process and response time latency/interference due to another process involved in attending to ambiguous (multiple response generating) stimuli.
  • On high-congruency Stroop tasks, schizophrenics show increased errors on incongruent relative to congruent trials, and increased facilitation on congruent relative to neutral trials. The implication being that in schizophrenics only one of the attention mechanism is selectively dysfunctional - that related to goal maintenance. As presumably, schizophrenics do not show abnormal patterns of reaction times (except for increased RT facilitation on congruent trails governed by the lack of maintenance of goal - 'ignore-the-color') , thus, the second mechanism involving selection of competing responses is intact.
  • ERP studies of Stroop tasks have identified a wave that may originate from anterior cingulate (ACC) and appears to correspond to response selection and competition processes; in contrast, the activity of a different wave up to 800 ms before stimulus presentation predicts correct performance on the next stimulus (and appears to originate from polar or dorsolateral frontal cortex [dlPFC]). The implication being that dissociated brain regions are involved in priming for the response (goal maintain ace) and selection of response ( conflict resolution - inhibition of inappropriate responses)
  • Event-related fMRI shows a strong negative correlation between delay-period dlPFC activity and Stroop interference, whereas ACC activity is tied to the presentation of incongruent stimuli. The implication being that PFC activity is related to errors and thus the process of goal maintenance, while ACC activity is related to peculiarities arising from incongruence - that is when competing responses are available- and thus tied to the process of response selection (inhibition of inappropriate response).

A clinching observation that could seal the argument about two dissociated mechanisms would be observing a correlation between errors on incongruent trials under 0 congruence condition (where the effect of goal maintenance has been effectively factored out by forcing subjects to keep the goal in mind on every trial), or better to display the goal (the rule that you have to choose as per the color and not the linguistic word) while the stimuli are presented to ensure that the goal is maintained constantly, and observe the correlations between errors in preceding condition and response time latencies/interference in the normal stroop task. This correlation would ensure that there indeed is an independent attentional mechanism that is independent of goal maintenance and is dependent only on conflict resolution.

In the second study by Fincham and Anderson, a learning paradigm was used whereby some sports names were associated with some arithmetical rules (that were either implicitly learned or explicitly told) and in the trials the subjects were required to retrieve the rule and apply it. There were four conditions - a visible-rule and rule-retrieval condition (supposed to measure the effects of the rule retrieval process) and a reverse/ forward calculation condition (supposed to measure the effect of rule complexity condition - a forward-reverse application of rule introduces another control step).

The authors discovered that in the first experiment, where there were four different trial conditions, recall (rule-retrieval condition) had a significant effect on both latencies and errors, they also found (but glossed over) a minor effect found of direction (or complexity of rule application) on the errors and latency and found no recall by direction interaction. Thus, it is evident that recall (or rule retrieval) and direction (or rule complexity/manipulation) are two different factors affecting performance. However imaging studies were not that helpful. Instead of finding a selective ACC activation effect linked to direction (as per their proposal of ACC as an attentional control region) and a selective PFC activation effect linked to recall (as per their proposal of PFC as a region involved in retrieval), they found that both recall and direction had effects on ACC and PFC activations.

Their second experiment was done with the purpose of dissociating the recall (retrieval) and direction (control) components. However they confounded the study by simultaneously introducing two variables- an additional direction task supposedly requiring an additional control step and not affecting retrieval at all, and a practice variable supposedly only affecting recall (retrieval) and not affecting control (rule manipulation) at all. This however cannot be taken for granted. All 3 trials in this experiment were recall trials. They present results for initial trial, a forward direction trial after some practice and a reverse direction trial after some practice. In my opinion, they should also have provided a simple direction-neutral trial after some practice. Comparison between this and the initial trial (which were same in all respects accept for practice) would have enabled a conclusive association of practice with retrieval ease and with decrease in PFC activation.

Even if the two practice trials (reverse + forward combined) are taken as a substitute for that direction- neutral practice trial (which they unfortunately did not conduct), still one can only derive the decrease in PFC activation due to practice (or ease of retrieval) relationship as a conclusion of this study. The increase in ACC that they observed between the initial trial and the final trials (involving reverse/ forward direction manipulation) are the same results that they observed in experiment one (whereby forward and reverse manipulations in both recall/ explicit condition led to more errors/ latency/ ACC activation) . They prefer to explain this as implying that ACC activation was required because an additional control step was involved); a more parsimonious ( and more in line with the current views of the functions of ACC) explanations is that when the reverse/ forward direction condition is added , then the stimuli that is presented (and which also contains the cue as to in which direction the calculation needs to be done) leads to a stroop-like default automatic forward direction application of the rule and ACC activity is required to choose between the competing responses (if reverse direction cue is present than forward direction response needs to be inhibited). This would predict more RT and errors in the reverse condition (incongruent trials) as opposed to forward conditions (congruent trials). One can even have some control conditions whereby novel sports words (with a novel explicit rule with no directionality associated with it) are displayed in some trials and reactions times and errors measured on these. If the resulting results are same as in Stroop task, perhaps the same mechanisms are in work.

The greater activation in ACC could also be, paradoxically, due to practice. To rule this out, the initial trials having both forward and reverse direction conditions should be compared with later reveres and forward direction trials after practice. Only if no increase in ACC activity is found that can be attributed to practice alone, can the increase in ACC be attributed to the additional control step that was supposedly introduced in experiment 2. A possible scenario where practice could influence ACC activation (and post stimulus response selection mechanism) is where practice or learning could lead to greater salience of activated goal or a stronger top-down expectation resulting in a stronger inhibitory signal for any stimulus that doesn’t meet the top-down expectations. It is not unreasonable to suppose that strength of a rule (the probability with which that rule has been ingrained in memory) may directly reflect in the strength of the biasing that is a result of a top-down expectation of that rule application. In that case , ACC may paradoxically be more and more activated as a result of practice (as the response expectation associated with the stimulus increases in habit strength though learning) to bias the response selection more strongly in favor of the expected response (goal).

In summation, there seems much ground to believe that two attentional processes in working memory /learning and performance are involved – one ACC based and the other PFC based and that they are explained in terms of pre-stimulus goal maintenance and post-stimulus response selection / biasing.

Sphere: Related Content

No comments: