Memory in Moment-to-Moment Action: Reactive Control in Older Adults

How does memory help to accomplish moment-to-moment goal-directed action? Classic accounts, such as Baddeley's working memory model, suggest that there are separate storage and processing ("executive") mechanisms, whereas newer accounts (proposed by a variety of researchers) propose that storage and processing are intertwined in the form of maintained goal or context representations.

According to these newer theories, individual differences in the strength of goal representations can more or less efficiently "bias" perception and behavior, particularly in cases where habit or environmental context might conflict with the inner goals.

In a new paper, Paxton, Barch, Racine and Braver elaborate the goal maintenance account to describe behavior among older and younger adults in a task that requires strong context representations - the AX-CPT.

People have a really hard time understanding this task (it is really complicated!). So rather than attempt to explain the task and methods first, as I normally do, here's the take home:

Older adults are impaired in their maintenance of context in a peculiar way. They attempt to compensate by recruiting more total brain volume, but are less precise in engaging regions thought to be important for goal or context maintenance (prefrontal cortex) when they need them most: during the delay-period of trials (when information needs to be maintained) and to cues (which need to be encoded for later maintenance). Instead, they show greater probe-related activity, as though they are attempting to reactivate the context representations which they had not successfully maintained. This is known as a "reactive" control strategy, and has not been often considered in any model of how memory guides moment-to-moment action (except, perhaps, in the field of prospective memory). This reactive control trend is particularly apparent in situations where context maintenance is most necessary for determining an appropriate response.

If you're curious about the details, I've summarized the most important ones below:

21 younger and 20 older adults completed the expectancy word AX-CPT, where one response is given to a particularly common word pair (FATE-LIME, occuring 70% of the time) and a different response to any other word pair (e.g., FATE-DOG, ARM-LIME, or ARM-DOG). All words are presented sequentially. Because of the higher frequency of the FATE-LIME pair, subjects with strong context representations will tend to anticipate one response when they see FATE, and may actually give the wrong response if a word other than LIME follows it. Conversely, subjects with weak context representations may give the wrong response when they see LIME, even if another word had preceded it. Each subject completed two versions of this task - one with a long delay (5 or 7.5sec) between words, and another with a shorter delay (1s) between words. The authors recorded accuracy, reaction times, and fMRI data from subjects completing this task. The authors z-transformed all RTs (within-subjects) to correct for individual differences in speed.
The results:
Older adults made significantly more mistakes than younger adults on trials where weak context maintenance can lead to an error (in the example above, "ARM-LIME") and numerically fewer mistakes than young adults on trials where strong context maintenance hurts ("FATE-DOG"). The reaction time results were analogous to these results, with older adults taking significantly longer on the trials where context maintenance is important, and responding significantly faster than young adults on trials where context maintenance is unhelpful. Furthermore, older adults' discrimination between contexts (calculated in terms of the signal detection measure d') changed as a function of delay more than the young adults' discrimination.
The neuroimaging results showed older adults recruiting much more total brain volume than younger adults, including right superior frontal gyrus, bilateral middle frontal gyrus, and right inferior frontal gyrus. However, younger adults showed more recruitment of right dorsolateral PFC and inferior frontal junction than older adults during the delay period (when no stimuli were present), suggesting younger adults were actively maintaining context during that time. Older adults still showed some recruitment of these and other regions (including left anterior cingulate and bilateral motor cortex).
The authors then conducted a second fMRI study of AXCPT, using another group of 16 younger and 16 older adults, but with an event-related design (which allows for better disambiguation of the time course of activity on each trial). In this design, the authors separated transient stimulus-related activity from sustained activity holding across trials, and further separation of activity due to the first or second stimulus in each stimulus pair (cues vs probes, respectively).
The behavioral results from the second study were slightly different, showing only that younger adults were more error prone than older adults only on the most frequent trials; both age groups were slower on trials where context maintenance is unhelpful relative to those where it helps.
In terms of sustained effects across the entire task, the neuroimaging results showed much greater recruitment of a variety of prefrontal regions by older than younger adults (consistent with the first study). However, when this activity was segmented into trials, the results reveal the reverse pattern: younger adults recruit prefrontal regions more strongly specifically during each trial. The authors mention that some regions showing decreased trial-related activity in older adults relative to younger adults are the same which also show increased sustained activity in older relative to younger adults.
The within-trial analyses showed that older adults engage less total brain volume to cues than younger adults, particularly in right superior, dorsolateral, and inferior frontal regions, as well as pre-SMA, while the reverse pattern was observed in left inferior frontal gyrus and anterior cingulate. This effect may have been driven by reduced activity in older adults (relative to younger adults) on the trials where the first word in a word pair was an infrequent (nontarget) word. Younger but not older adults showed enhanced activity on those trials relative to when the first word was a frequent (target) word.
This general pattern was reversed when looking at the second word in each pair, such that older adults generally recruited more brain volume at that time than younger adults, particularly in right dorsolateral PFC and middle frontal gyrus, as well as left superior frontal gyrus. This was particularly prominent on trials where context maintenance is important (where the first word is infrequent but the second is most frequently paired with a target response).

Related Posts:
Effects of Aging and Delayed Execution on Prospective Memory
Localizing Executive Functions in Prefrontal Cortex (left.IFG as involved in "reactive suppression")


More like this

As described in yesterday's post, many theories have been proposed on the possible functional organization of prefrontal cortex (PFC). Although it's clear that this region plays a large role in human intelligence, it is unclear exactly "how" it does so. Nonetheless at least some general…
"Instead of trying to produce a programme to simulate the adult mind, why not rather try to produce one which simulates the child's?" - Alan Turing (Computing Machinery, p456) One of the defining features of childhood cognition is "behaving without thinking." Not surprisingly, developmental…
When I started this blog back in '06, new hypotheses were appearing on a possible functional architecture of the lateral prefrontal cortex - a recently-evolved brain area implicated in high-level cognitive functions like planning, analogical reasoning, and cognitive control. Since then, these…
Ideally, our real-world behavior is strongly determined by our context, for the simple reason that some behaviors are only appropriate in some situations (e.g., eating during an internal context of hunger, or using slang during an external context of casual interaction). Context-inappropriate…

Thanks for the summary. For those interested in more of Braver's work, I highly recommend his chapter in Variation in Working Memory. Also, I don't think the AX-CPT task is that complicated. That is part of what attracts me to this work, I think this task makes context processing easy to understand.

I love the ViWM chapter as well (though I think there's a lot missing from the current reactive control account - exciting work ahead I think).

Do you work on AX-CPT? I'm finishing up a project with AX-CPT in 3 & 8 yr olds - hopefully submitting soon...

No, I don't currently work with AX-CPT. Good luck with your submission.