A new way to analyze human reaction times shows that the brain processes data no faster than 60 bits per second.
For more than a century, psychologists have used reaction time as a window into the brain. The thinking is that information processing takes time, so the average time taken to begin or complete a task reflects the duration of the cognitive processes involved in it.
For example, a typical reaction time experiment might ask a subject to classify a sequence of letters as a word or a nonword, by pressing a button. This kind of experiment is called a visual lexical decision task.
This information-centric approach is clearly ripe for an information-theoretic treatment. And sure enough, no sooner had Shannon published his theory of information in the 1940s, psychologists began to apply it to the exchange of information between the environment and the brain that goes on during reaction time experiments.
Their approach eventually lead to Hick's Law, one of the few laws of experimental psychology. It states that the time it takes to make a choice is linearly related to the entropy of the possible alternatives. The results from various reaction time experiments seem to show that this is the case. Although one by-product of this approach is that the results are intimately linked to the type of experiment used to measure the reaction time. And that makes each study peculiarly vulnerable to the idiosyncracies of the experimental approach.
Today, Fermín Moscoso Del Prado Martín from the Universite de Provence in France proposes a new way to study reaction times by analysing the entropy of their distribution, rather in the manner of thermodynamics
The entropy is an estimate of the amount of information needed to specify the state of the system.
Martin says the the entropy of the distribution of reaction times is independent of the type of experiment and so provides a better measure of the cognitive processes involved. That's important, not least because it provides a way to more easily compare the results from different types of experiment.
Martin uses his method to determine how much information the brain can process during lexical decision tasks. The answer? No more than about 60 bits per second. Of course, this is not the information processing capacity of the entire brain but one measure of the input/output capacity during a specific task.
Martin goes on to analyse the data from various types of reaction time experiment, in particular to determine whether information processing speed is constant during a particular task, as implied by Hick's law. Martin reckons it isn't.
"This finding suggests an adaptive system where the processing load is dynamically adjusted to the task demands," he says. That makes sense. It seems crazy to assume that the brain carries on processing data at the same rate regardless of the complexity of task at hand.
But this has an important implication: that the linearity of Hick's law doesn't always apply. So Hick's law will need some kind of modification to cope with this non-linearity.
Just how to re-write one of the basic laws of behavioural psychology isn't clear yet. But it's sure to involve a very different way of looking at the brain form when it was formulated.
at the Technology review by MIT
No comments:
Post a Comment