Can cognitive processes be inferred from neuroimaging data?

Abstract

There is much interest currently in using functional neuroimaging techniques to understand better the nature of cognition. One particular practice that has become common is 'reverse inference', by which the engagement of a particular cognitive process is inferred from the activation of a particular brain region. Such inferences are not deductively valid, but can still provide some information. Using a Bayesian analysis of the BrainMap neuroimaging database, I characterize the amount of additional evidence in favor of the engagement of a cognitive process that can be offered by a reverse inference. Its usefulness is particularly limited by the selectivity of activation in the region of interest. I argue that cognitive neuroscientists should be circumspect in the use of reverse inference, particularly when selectivity of the region in question cannot be established or is known to be weak.

Role of the anterior cingulate and medial orbitofrontal cortex in processing drug cues in cocaine addiction

“…Thus, although -by design (see Supplementary Material) -mean performance did not differ between the two task conditions, erroneous performance correlated with hypoactivation in the rACC/mOFC and this effect was context specific (i.e., observed only when subtracting neutral from drug responses) ( Figure 1C); and (2) the more the cdACC activation specifically to the drug (and not neutral) words, the more negative were the valence ratings attributed to the drug (and not neutral) words ( Figure 1D). Note that our subsequent interpretations are therefore based on these significant brain-behavior correlations, not to be attributed to 'reverse-inference' (Poldrack, 2006).…”

Neural evidence for “intuitive prosecution”: The use of mental state information for negative moral verdicts

“…Within the ROI, the average percent signal change (PSC) relative to fixation (PSC = 100 × raw BOLD magnitude for (condition − fixation)/raw BOLD magnitude for fixation) was calculated for each condition at each time point (averaging across all voxels in the ROI and all blocks of the same condition). PSC during story presentation (adjusted for hemodynamic lag) in each of the ROIs was compared across experimental conditions (Poldrack, 2006).…”

“…extra mental state processing). The validity of a reverse inference depends on the prior evidence of the target brain region's selectivity for the cognitive process in question (Poldrack, 2006;Young & Saxe, 2009a). Of the regions implicated in theory of mind, the RTPJ appears to be especially selective for processing mental states such as beliefs, in and outside the moral domain Saxe & Powell, 2006 ;.…”

Neural correlates without reduction: the case of the critical period

“…The search for neural correlates is often tied to a "localization strategy," a research program that posits that activation in a neural structure is "maximally sensitive" or "selective" for a specific psychological function. This posit is what enables the practice of "reverse inference" in neuroscience, which concludes that a particular psychological function is engaged on the grounds that a certain neural structure is activated (Poldrack 2006). To be sure, localization can allow that there can be more than one neural correlate for any given psychological function, in other words, that co-activation of multiple structures is necessary for the implementation of a certain psychological function.…”