Memory strength and specificity revealed by pupillometry

Voice-specificity effects in recognition memory were investigated using both behavioral data and pupillometry. Volunteers initially heard spoken words and nonwords in two voices; they later provided confidence-based old/new classifications to items presented in their original voices, changed (but familiar) voices, or entirely new voices. Recognition was more accurate for old-voice items, replicating prior research. Pupillometry was used to gauge cognitive demand during both encoding and testing: enlarged pupils revealed that participants devoted greater effort to encoding items that were subsequently recognized. Further, pupil responses were sensitive to the cue match between encoding and retrieval voices, as well as memory strength. Strong memories, and those with the closest encoding-retrieval voice matches, resulted in the highest peak pupil diameters. The results are discussed with respect to episodic memory models and Whittlesea's (1997) SCAPE framework for recognition memory. ► We examined recognition memory for spoken words while tracking pupillary changes. ► Words were heard in different voices; half changed in the recognition test. ► Pupil dilation during learning predicted later memory accuracy and confidence. ► Study-to-test voice changes were reflected in pupil dilation during test. ► Pupillometry reveals cognitive effort during memory encoding and retrieval.