It is rare that we can do two things at once—if someone asks you to rub your stomach and pat your head, odds are that you’ll rub your head. A better strategy for a lot of different cognitive tasks is to switch between them. Switching typically comes at a cost, however. Switching can even negatively impact skills that we are experts in. It is clear that using a phone or other screen device while driving poses a danger, and driving makes us worse conversation partners.
We can assess these costs using classical experimental paradigms in attention research, such as the “psychological refractory period.” In these paradigms, participants see two stimuli A and B and then have to make two separate responses – one for each stimulus. The responses can be similar or different, such as whether a shape A is orange or blue and whether a spoken word B starts with a “k” or not. Because participants have to make two decisions, one response to A must be completed before the response to B can begin. The general pattern is that the closer in time A and B appear, the longer participants take to respond to B, with no effect on A. Responding to A imposes a bottleneck on subsequent processing.
A major open question in this area of research is, what determines what gets done first? Can we predict or manipulate which task goes through the bottleneck first?
Classic studies in this area have typically tested whether the information that is presented first goes through the bottleneck first, in what might be called a “first-come, first-served” account. This has been tested by stretching out the amount of time that A is on the screen. If the shape A stays on the screen until after B starts playing, B may go through the bottleneck first. In this case participants could make a judgment about B (say, with the right hand) before they make a judgment about A (with the left hand).
Another running theme is that researchers often tell participants to respond to A and B as they arrive, which could influence whether participants respond to A and B in that order. We can think about these different configurations as in Figure 2 below.
In a recent article published in the Psychonomic Society journal Attention, Perception, & Psychophysics, a group of researchers led by Tilo Strobach manipulated (a) differences in instructions and (b) the amount of time a stimulus lingered on the screen. Their design allowed them to test what influences which task is responded to first. Their study is one of the first to tease out the “first-come first-served” principle from the order of presentation and instructions. The authors conducted two experiments, but we will focus on Experiment 1.
Strobach and colleagues used a familiar “dual task” paradigm in which participants did two things at once. The first task was a visual discrimination task, in which participants specified the number that was on the screen. The second was an auditory discrimination task, in which participants classified tones into “high”, “medium”, or “low” categories.
To test the idea that slowed perceptual processing can lead to task switching, Strobach and colleagues created two versions of the visual task, corresponding to strong-contrast and weak-contrast conditions. Decreasing the contrast between the number and the background should make perceptual processing more difficult and increase the amount of time required before a motor response can be planned. This should lead to more task switching.
Strobach and colleagues also wanted to know how much it mattered that participants were asked in previous studies to respond “in the order of appearance.” So, half of the participants were allowed to respond according to their preference, and the other half were told to respond in order.
The major manipulation in all studies like this one requires manipulating the lag between the events A and B, known as stimulus onset asynchrony, or SOA. The greater the SOA is between A and B, the less of an influence A has on B. The final manipulation in this experiment was whether the visual task preceded the auditory task, and by what length of time. The experiment included SOAs of 0, 60, 120, and 400 millisecond.
The authors use an interesting trick to plot the two presentation orders (AB or BA) by making the SOAs negative when the numbers were presented before the tones. In keeping with the “first-come, first-served” principle, whichever task starts first should influence which response participants make first.
This general pattern is what they found, consistent with the literature. The earlier the tone plays before the numbers appear on the screen, the more likely participants were to respond to the tone first. What is particularly interesting is that making the visual display harder to process by reducing the contrast between the number and the background, is that it changed the order in which participants responded. Once it was harder to see the exact number, participants started responding to the tone first. As predicted, providing instructions that allowed participants to respond as they wanted to the two different tasks chipped away at this tendency, though the same general pattern still emerged – when one decision is hard to make because it is hard to see, the other decision becomes relatively easier to make. This provides some evidence for the first-come, first-served principle.
Despite the clear evidence that participants sometimes switched to the tone-discrimination task when the visual discrimination task got more difficult, it is less clear what happens when participants are told to respond in a given order. When having to keep a specific order in mind, this can add some additional processing costs, making a task more difficult. So, the authors looked at the reaction times for both tasks for the decisions, which can sometimes tell us a different story.
Earlier we pointed out that the bottleneck typically manifests in reaction times to the second response (B), because the first (A) tends to be unaffected by experimental manipulations like the lag between tasks (SOA). Strobach and colleagues found a very interesting pattern of results. For the first responses (which could be either visual or auditory), there was indeed no relationship between SOA and reaction time. However, reaction times increased consistently with respect to SOA for the second response. And finally, if participants had to respond in a particular order (the order of arrival), there was an additional cost imposed across the board, but only if the visual display was degraded (i.e., the weak contrast condition). There were no general differences if participants could decide what to do and when.
The results of this study make it clear that the task that comes through the bottleneck first gets preference when we are multitasking, especially under adverse conditions. While there are differences in response patterns when participants are allowed to choose order of responding, this does not completely undo the “first-come, first-served” principle. Instructions influenced reaction times more clearly than they influenced response order, but in a way that suggests that the additional task of having to remember what came first imposes a global cost and slows decisions down.
Researchers can breathe a sigh of relief knowing that specific order instructions seem to just have a global effect on processing time. Finally, in what is sure to be good news for participants in the lab, the results of Strobach and colleagues’ study suggest that giving participants a bit more autonomy during the decision-making stage can lead to the same patterns of results as when they are given specific instructions.
Psychonomics article focused on in this post:
Strobach, T., Hendrich, E., Kübler, S., Müller, H., & Schubert, T. (2018). Processing order in dual-task situations: The “first-come, first-served” principle and the impact of task order instructions. Attention, Perception, & Psychophysics, 80, 1785-1803. DOI: 10.3758/s13414-018-1541-8.