Groundhog Day is better for your homework

Groundhog Day is better for your homework: We adapt to attentional conflict but only if nothing changes

William James famously postulated that the world presents itself as “one great blooming, buzzing confusion” to an infant, whose senses are constantly assaulted by visual, auditory, tactile and olfactory stimuli. To make sense of the world requires attention, the ability to focus on one stimulus in favor of another, even if the TV is running while doing the homework.

This attentional task becomes more difficult when there is conflict between aspects of stimuli that needs to be resolved. The prime example of such conflict is the Stroop task, which we have discussed here previously. In the Stroop task, people have to name the color ink in which a word is printed. Although this sounds easy, competition can be easily introduced by presenting the word “red” in blue ink. The required response is “blue” but it is difficult to suppress the immediate temptation to say “red” and so people are slowed considerably on those incongruent trials compared to congruent trials (“red” in red ink).

How do we cope with the competition during incongruent trials?

According to the conflict-monitoring hypothesis, conflict detection in a current incongruent trial will invoke cognitive control processes, such as an attempt to increase the weight given to the relevant features (color rather than meaning). In consequence, on the next trial, the speed disadvantage of incongruent trials relative to congruent trials is reduced.

In other words, saying “green” to the word “yellow” presented in green ink is easier after a trial involving “red” presented in blue than a trial involving “red” presented in red. We seem to adapt to the conflict and get better at resolving it on the next trial.

Intriguingly, however, there is evidence that this adaptation is task specific. That is, resolving conflict in a Stroop task does not help us if the next trial involves another sort of conflict—for example a Simon task. In a Simon task people make a choice about a stimulus (e.g., “is the square red or green?”) and indicate their response by pressing a key with one hand or the other (e.g., left hand for red, right hand for green). The crucial manipulation is the compatibility between the location of the stimulus and the associated response. The Simon effect refers to the observation that when the stimulus and response locations are compatible (red square on the left or green on the right), people respond more quickly than when the locations are incompatible (red square on the right, or green on the left).

recent article in the Psychonomic Bulletin & Review sought to explain the presence as well as absence of adaption to conflict by invoking a unifying principle based on how continuity (or change) of context signals the need to maintain (or discard) previous task-relevant cognitive settings. When the processing context changes, the continuity of cognitive processing across trials is disrupted, thereby abolishing the benefits of adaptation to conflict that is observed when context is invariant.

Researchers Kreutzfeldt, Stephan, Willmes and Koch tested this hypothesis in an experiment in which tasks remained constant across trials, but the modality of the task (visual vs. auditory) could alter between trials. If context matters, then adaptation should be disrupted by a change in modality even though the task remained the same throughout.

Kreutzfeldt and colleagues presented participants with two tasks, each of which was presented both visually and auditorially. In the location-judgment task, a tone was played through the left or right channel of headphones, and a diamond symbol appeared to the left or the right of the center of the screen. In the numerical judgment tasks, participants had to identify which of two numbers (2 and 8) was presented by headphone and on the screen. In both tasks, participants used two keys, on the right and the left of the keyboard, to record their response.

The two tasks were “blocked” so that participants could anticipate the type of task they had to perform. However, the modality of the task was varied from trial to trial, and was indicated before each trial commenced. Thus, when participants heard a tone before a trial they knew to attend to the information presented via headphones and ignore the screen. When they saw an “x” on the screen instead, they knew to attend to the screen and ignore the headphones information.

The first crucial manipulation was whether the visual and auditory modalities were congruent or incongruent. On congruent trials, the stimulus tone might be presented on the left (for the location task) and the diamond would then also appear on the left. On incongruent trials, the tone might still be presented on the left but the diamond would appear on the right. (Congruence was manipulated in an analogue fashion for the numerical judgment task.) Knowing which modality to attend to was therefore crucial to perform the task. And of course, the incongruent trials induced considerable conflict, similar to “red” being presented in blue ink.

The second crucial variable involved the pairing of contiguous trials: modality could either be repeated or switched between pairs of trials, and each trial in a pair could either be congruent or incongruent. Would the adaptation that is typically observed after the first incongruent trial—such that the next incongruent trial incurs less processing cost—persist when modality is switched?

The results are shown in the figure below, which plots the “congruency effect” observed on the second trial in each pair as a function of task, modality transition, and the congruency of the preceding, first trial in a pair (called n-1 in the figure). The congruency effect is defined as the difference in response time between incongruent and congruent second trials in a pair.

The figure shows several interesting patterns. First, a congruency effect is obtained in all conditions—all of the means are positive, implying that an incongruent second trial in a pair slows down performance compared to a trial in which both modalities indicate the same response. This overall pattern establishes the presence of attentional conflict when the two modalities are incongruent.

Second, when modality was the same on both trials within a pair (left pair of bars in each panel labeled “repetition”), the congruency effect was much reduced if the preceding trial was also incongruent (white bars), compared to when it was congruent (gray bars). This effect replicates the adaptation to incongruency that has been observed in the previous literature: having battled competing stimuli once makes it easier to do so a second time.

The third result is of greatest interest because it is novel: In both tasks, the adaptation advantage arising from a preceding incongruent trial is eliminated if modality is switched. That is, the grey and white bars do not differ for the “switch” pair in each panel (if anything, for the location task the difference points the other way, although not significantly so).

The results thus support the idea that cognitive adaptation to conflict is highly specific, and limited to the particular modality required by a task. If that modality shifts, then even though the task remains unchanged, subsequent trials do not benefit from adaptation. The authors suggest that adaptation may require the presence of identical task-relevant features in working memory from trial to trial. If an “attentional reset” is required because modality changes, then working memory has to be reloaded with a new set of relevant features, thereby eliminating the lingering benefit of a previous trial.

So if your teenagers insist on watching TV while doing the homework, at least make sure that they watch Goundhog Day.

Article focused on in this post:

Kreutzfeldt, M., Stephan, D. N., Willmes, K., & Koch, I. (2016). Shifts in target modality cause attentional reset: Evidence from sequential modulation of crossmodal congruency effects. Psychonomic Bulletin & Review. DOI: 10.3758/s13423-016-1001-1.

You may also like