Your IQ can be reliably predicted by simple reaction time tasks – perhaps even more reliably than with much more complex cognitive tasks. This surprising psychometric fact has led to the belief in human “processing speed.” In the same way that a computer with a faster microprocessor might carry out more computations, with potentially less demand on memory, the idea is that brains with better neuronal efficiency also manifest both higher IQ and proportionately faster reaction times even in simple tasks.
To me, this story always seemed “too good to be true” – or perhaps merely “too simple to be true” – and a recent paper confirms this intuition (big hat tip to Kevin McGrew). The new paper, by Helmbold, Troche and Rammsayer, indicates that “temporal acuity” may be the reason that such simple reaction time tasks correspond with higher level cognition: the temporal resolution of neuronal processing, rather than its absolute speed, may be the factor that matters.
Previous work had identified that subjects’ ability in fine-grained temporal tasks (temporal order judgement across just a few milliseconds; auditory flutter fusion [the point at which a pulsed sound is perceived as continuous], and two-tone discrimination over short durations) predicts general intelligence better than the simple reaction time tasks. The authors confirmed this idea by administering many tasks (described individually here) to 260 adults, aged 18 to 39 years, with a broad range of education levels. The tasks fall into 3 broad categories: traditional intelligence assessments, reaction time tasks, and temporal acuity tasks.
The authors analyzed the results by extracting variation in each set of tasks that was consistent within individuals (known as factor analysis) to arrive at three factors: a general intelligence factor, a reaction time factor, and a temporal acuity factor. The authors then attempted to fit various particular theoretical assumptions to the data (known as structural equation modeling).
The reaction time and temporal acuity factors were significantly better at predicting general intelligence if they were allowed to correlate with one another, suggesting important overlap in these measurements. Further analyses indicated that only the temporal acuity factor had additional, independent predictive value that went beyond this overlapping variance.
The authors suggest that temporal acuity is incorporated in the reaction time measurements, and that fact accounts for their shared relationship with general intelligence. The more important and predictive variable is the cognitive ability measured by temporal acuity tasks, which they take to reflect the rate of neuronal oscillations.
So, faster oscillations = faster reaction time = better temporal discrimination.
I think this is an interesting (but probably wrong) idea, for the following reasons:
1) As the authors acknowledge, they only used the Hick paradigm to measure information processing speed, meaning that the factor extracted from this task might include a lot of uninteresting task-specific (as opposed to “cognition-specific”) variance.
2) I was under the impression that Raven’s matrices are the gold standard fluid IQ test – but these weren’t included here. So I wonder about the validity of their general intelligence factor.
3) The authors are making claims about neuronal processing based on behavioral data. Although behavioral data isn’t completely opaque with respect to mechanisms, there are many other more obvious explanations which go unmentioned (but which I describe below).
4) The most obvious way of testing the idea that neuronal oscillation rate predicts intelligence is to use data from intracranial or even EEG recordings, and thus to directly quantify the neuronal oscillations which supposedly correlate with general intelligence. Although such data is plentiful, no one (to my knowledge) has ever reported such a correlation.
5) Most of the temporal tasks all involve “oddballs” – stimuli which are contextually infrequent or otherwise “deviant” from the adjacent stimuli. Oddball detection may be an important individual difference in its own right, and I’m not sure temporal acuity would predict general intelligence so strongly if this task characteristic wasn’t so strongly represented in the tasks.
There is probably some truth to the idea that the temporal coherence of neural processing has potentially ubiquitous effects on behavior, in much the same way that general intelligence is the manifestation of a consistent, ubiquitous individual difference variable.
But “temporal acuity” is probably itself highly multifactorial, dependent on the precise regions of the brain most involved in a particular task and how efficiently they can interact to support performance. Still, temporal processing is a very promising – and often ignored – approach for determining how the brain gives rise to general intelligence.