Accurate processing of facial displays of emotion is critical for effective communication. A robust literature has documented impairment in the ability to recognize facial affect in people with traumatic brain injury (TBI), but research is scarce about memory for facial affect. Disruptions in recognizing and remembering the emotions of others can undermine relationship quality and may result in psychosocial dysfunction. Importantly, the extant literature indicates that facial affect recognition dissociates from other cognitive abilities such that it is likely a distinct neuronal process. Thus, explicit measurement of affect recognition and memory for emotions may be critical for implementing and refining rehabilitation interventions. The present study examined the relationship between recognition and memory for emotions using a novel computerized task and explored its associations with other cognitive abilities.

Participants were adults who were neurologically healthy (n = 31) or had a history of moderate to severe TBI (n = 26). The battery included the novel Assessment of Facial Affect Recognition and Memory (AFARM), Cambridge Face Memory Test (face memory without emotion), Wechsler Test of Adult Reading, Rey Auditory Verbal Learning Test, Judgment of Line Orientation, Oral Symbol Digit Modalities, Digit Span, FAS, Animal Fluency, and the Affect Intensity Measure (experienced emotion). Spearman correlations examined the relationship of AFARM performance with the test battery. Logistic regression models examined whether immediate-delay (ID-EM) and long-delay face emotion-memory (LD-EM) accounted for unique variance in group membership beyond recognition accuracy of facial affect and memory for faces.

AFARM demonstrated relationships with neuropsychological and mood variables in the expected directions across and within groups, with the strongest associations observed for memory for verbal information (rs = .51 to .58) and processing speed (rs = .48 to .57). Consistent with traditional list-learning tests, ID- and LD-EM were highly correlated (r = .85). Experienced affect intensity was inversely associated with ID-EM (r = -.29) and LD-EM (r = -.38) but not with recognition accuracy (r = -.10). Logistic regression examining ID-EM was significant, χ2(3) = 26.05, p < .001, Nagelkerke R2 = .49. ID-EM accounted for unique variance in group status (p = .006; OR = 0.65) after accounting for recognition accuracy and face memory. Similarly, the model examining LD-EM was significant χ2(3) = 27.70, p < .001, Nagelkerke R2 = .43; LD-EM was significant after accounting for other variables (p = .017; OR = 0.69).

The findings are consistent with the hypothesis that memory for emotions represents a unique component of social cognition that is separate from recognition. Accuracy in identifying emotions, face recognition memory, and memory for emotions are strongly related but not wholly redundant processes. Consistent with prior literature, subjective experience of emotion had substantial effects on objective performance tasks, indicating that an individual's intense experience of their own emotions can disrupt sensitivity to the emotions of others. Future research should assess the extent to which memory for emotions relates to psychosocial outcomes such as the quality and quantity of interpersonal relationships.