The start of a Formula 1 Grand Prix is always exciting and adrenalin producing, even if you watch it on TV from thousands of miles away and keep the noise level below the pain threshold. (A Formula 1 cockpit is one of the loudest places on Earth.)
Have a look at a start of a Grand Prix here:
You can calm down, no one got hurt.
Did you notice how everything got underway when the set of 5 red lights above the pack of cars were extinguished? Those red lights are of considerable cognitive interest. The starting regulations for Formula 1 specify how they operate: They illuminate one at a time, from left to right, at one-second intervals. They then go out simultaneously after an interval of between four and seven seconds. When the lights are extinguished, the race begins.
This is a classic temporal preparation task.
In a temporal preparation paradigm there is a warning stimulus (S1) and a target stimulus (S2). The reaction time (RT) in response to S2 is of greatest interest. In the case of Formula 1, S1 is the onset of the final light, and S2 is the offset of all the lights together. There is a large body of research showing that RT is influenced by the “foreperiod”, that is the duration and distribution of the time intervals between S1 and S2.
When the foreperiod is variable and drawn from a uniform distribution on each trial, RTs tend to decrease as the foreperiod increases. The longer you have waited, the more likely it is that you will stop waiting and hence you’re faster to get going. This decrease of RT becomes even steeper if the foreperiods follow an “antiexponential” distribution; that is, a distribution with few short foreperiods and many long ones.
By contrast, the RT-foreperiod function is approximately flat when the foreperiods are distributed exponentially. In that case, most foreperiods are short, few are long, and hence you are preparing yourself to get going from the onset of the foreperiod.
These empirical regularities are well described by the so-called hazard function of the distributions of foreperiods. The hazard function describes the probability of an event occurring in the next time interval, given that it has not yet occurred. If the hazard function is flat across time (as for an exponential distribution) then people’s RT is also unaffected by how much time has lapsed. If the hazard function increase (as it does for a uniform distribution) then people’s RT also increases with increasing foreperiods.
By the way, I was unable to ascertain what distribution is used in Formula 1.
The appeal to hazard functions is, however, cognitively somewhat unsatisfactory. While the hazard function may describe what people do as a first approximation, it does not explain anything. So how do we explain why people track the hazard function? What cognitive process makes us respond more quickly if a foreperiod sampled form a uniform distribution is particularly long? Why do we not speed up if the same foreperiod was sampled from an exponential distribution?
Researchers Rozemarijn Mattiesing, Wouter Kruijne, Martijn Meeter, and Sander Los recently published an article in the Psychonomic Bulletin & Review that continued their program of exploration of the cognition underlying human temporal preparation. They tested the multiple trace theory of temporal preparation (MTP) which proposes that temporal preparation is driven by the retrieval of memory traces of past experiences from long-term memory, rather than by knowledge about upcoming events.
The basic prediction of this theory is that when the distribution of foreperiods changes across trials, people’s temporal preparation will continue to be affected by their prior experiences. That is, our memory for the experienced foreperiods will determine how we prepare ourselves for the next trial.
Mattiesing and colleagues tested this prediction in an experiment that spanned two sessions that were scheduled a week apart. The figure below shows the design.
The crucial manipulation consisted of the distribution of foreperiods that participants experience in the first session. For one group the foreperiods were mainly exponential, and for the other group they were anti-exponential. The question of interest was what would happen during the second session in which the foreperiods were all uniform.
Each trial commenced with presentation of S1, a black cross (+) in the center of the screen. Then, after a variable foreperiod as determined by the distribution for the particular group and block, S2 (a black square) was presented. Participants had to determine as quickly as possible whether S2 was to the right or left of the center of the screen (it was randomly offset by a small amount).
The results are shown in the next figure.
It is immediately apparent that during the first session (first 5 panels), participants were sensitive to the distribution of foreperiods, as would be expected from the previous literature. Participants in the anti-exponential condition (open circles) showed an extremely steep decline in RT as the foreperiod increased, whereas those in the exponential condition (filled circles) remained more or less flat.
Of greater interest is that this effect persisted, albeit in an attenuated manner, in the second session a week later. People’s prior experience still affected the way they were preparing for the task, even after multiple blocks of practice on the new, uniform distribution of foreperiods. It was only in the final block (rightmost panel) that the interaction between Group and foreperiod was no longer significant—up to then, prior experience from a week ago still mattered.
These data provide strong evidence that people form surprisingly lasting memories for the distribution of preparation times that they experienced a week previously. This effect is difficult to reconcile with any model that proposes that people estimate the hazard function for the task at hand in an ahistorical manner.
Memory matters a great deal to what people do, even in situations where it is not explicitly required.
Article focused on in this post:
Mattiesing, R. M., Kruijne, W., Meeter, M., & Los, S. A. (2017). Timing a week later: The role of long-term memory in temporal preparation. Psychonomic Bulletin & Review, 24, 1900-1905. DOI: 10.3758/s13423-017-1270-3.
