Abstract
There is growing interest in using reaction time (RT) during recognition tasks as an index of face (or object) recognition ability. However, whether RTs are valid in this context and whether they explain recognition ability variance above and beyond accuracy alone has not yet been established. Decades of individual differences research in other domains, particularly in intelligence testing, consistently shows that faster RTs on very simple RT tasks are predictive of greater cognitive ability. This robust association has yet to be accounted for theoretically and further, it is not clear whether this association holds when using RTs from more complex tasks. To better characterize the RT/ability relationship during simple and complex tasks, the present study (N=2,627 participants) examined the associations of RTs on the simpler vs. more cognitively demanding stages of the Cambridge Face Memory Test (CFMT) with face recognition ability as independently defined by Famous Faces Memory Test (FFMT) score. We found that CFMT section RTs explained ~5% additional variance in FFMT score beyond CFMT section accuracies. Importantly, the direction and presence of the correlation between CFMT RTs and FFMT scores differed by CFMT section. Specifically, RTs from the simpler, delayed-match-to-sample section negatively correlated with FFMT ability level. RTs from the intermediate section did not correlate with ability level despite our massive sample, but RTs from the most demanding section, involving recognizing multiple faces from foils in visual noise, positively correlated with ability. CFMT trial accuracies indicated that waning motivation for lowest ability participants did not account for this shift. These results reveal that task RTs may be useful as an index of face recognition ability, and, more generally, suggest that faster RTs do not universally predict greater cognitive ability. Rather, task demands can modulate the direction and magnitude of the relationship between RT and cognitive ability.
Meeting abstract presented at VSS 2018