Developing Intelligence

The distributions of reaction times are always positively skewed, which seems to reflect two independent processes: a normal gaussian distribution of reaction times, in addition to an exponentially-decaying distribution of a few very long trials. Measures of this reaction time (RT) variability show surprisingly strong correlations with fluid intelligence, and have been interpreted to reflect things as diverse as goal neglect, overly liberal updating of working memory, erratic bursts of dopamine release, or tonically elevated levels of noradrenaline. But these explanations are not mutually exclusive.

In fact, they are incredibly complementary: updating seems to mediate the correlation between tests of executive function and IQ, is frequently understood as resulting from phasic dopamine release, and phasic dopamine release itself may be more likely when noradrenaline levels are relatively high.

This makes reaction time variability a fascinating measure, although there is understandable controversy on how to interpret it. Unfortunately, the common practice of log-transforming reaction times obscures this variability, and the common practice of simply taking the median RT confounds these two sources of variance (although it’s certainly better than using mean RT).

However you measure it, RT variability may be particularly important in putative measures of inhibition. Some have suggested that cerebellar and noradrenaline malfunction (both of which are involved in reaction time accuracy) may mediate the correlation between ADHD and impaired response inhibition. An in-press article in Neuropsychologia from Simmonds et al. directly addresses this issue by imaging the neural correlates of response variability in response inhibition tasks among children.

30 healthy children, aged 8-12 years, were presented with images of either green or red spaceships while in the fMRI scanner. Subjects had to respond as quickly as possible to green spaceships, but not to respond at all to red ones; green spaceships occurred 75% of the time. Reaction time variability on Go trials was measured as the ratio of RT standard deviation to RT mean.

The results showed, as you might expect, that children with greater RT variability had more errors of commission – mistakenly giving a response to the NoGo “red spaceship” stimuli. Over all participants, RT variabililty did not correlate with full-scale IQ; but among the top and bottom 8 subjects in terms of RT variability, IQ did significantly correlate, consistent with previous literature.

Successful inhibition was associated with increased activity in pre-supplementary motor area, and that was in turn related to lower variability in RTs. In contrast, higher RT variability was associated with less anterior cerebellum and more dlPFC activity generally, more dlPFC & AFC activity in NoGo inhibition trials, and more posterior cerebellum activity in Go trials.

These results are generally consistent with a view of the Go/NoGo task in which commission errors are caused by erratic patterns of reaction time rather than deficits in response inhibition per se. Variable RTs may emerge from a failure to properly recruit the cerebellum, and may be accompanied by compensatory activation of prefrontal regions. In contrast to previous evidence, successful inhibition in this task was not associated with right vlPFC activation in any of the contrasts here.

One aspect of this study clashes with previous theoretical accounts of inhibition. According to the influential “Race model,” two distinct and competing processes interact to produce performance in these tasks: a “go” process and a “stop” process. This account is partially supported by the lack of previous correlations between “stop signal reaction time” (theoretically, the amount of time for the “stop” process to execute) and reaction time on “Go” trials. This article demonstrates that these two processes are not completely independent, as reaction time variability (necessarily on “Go” trials) correlates with comission errors.

However, this work accords with recent research on the role of the subthalamic nucleus (STN) in putative response inhibition tasks. According to this paper, the STN inhibits bursting activity throughout the striatal Go pathway (thought to be involved in executing motor behaviors and possibly also in updating working memory), allowing more information to be “integrated” before the execution of a response. STN receives direct projections from anterior cingulate, suggesting that STN activity may underlie classic speed-accuracy tradeoffs. In this case, RT variability could also result from impaired STN function.

Related Posts:
fMRI of the Stop Signal Task: What Computations Support “Stopping”?
Sensitivity to Frequency: A New Model of Hemispheric Asymmetry
Dissociable Inhibition Mechanisms In Motor and Oculomotor Tasks
Stop vs. Go: Re-Examining the Race Horse Model of Inhibition
Exploration & Exploitation Balanced by Norepinephrine & Dopamine
Tonic Dopamine and Response Variability
Dopamine for Dummies