Event-Related Potential and Behavioral Correlates of Deafness to Change in the Identity and Location of Objects in Auditory Scenes

Abstract

ABSTRACTFailure to detect change in an auditory scene (i.e., change deafness) has been found to occur with alterations to object identities and the location of objects. Event-related potential (ERP) correlates of change deafness for object identity have been identified, but to our knowledge, ERP comparisons between spatial and identity change deafness are non-existent. We examined whether ERP differences exist between spatial and identity (ID) changes, and among levels of change detection accuracy (correct change identified vs. wrong change identified vs. change deaf). Within a trial, listeners were presented with two consecutive auditory scenes, each composed of four environmental sounds played simultaneously at separate azimuths on the horizontal plane. Scene 2 was either identical to the first (“no change”), had one sound replaced (“ID change”), or contained the same sounds at different locations (“space change”). Accuracy was similar for ID and space change trials (~60% correct), but error patterns differed. On space change trials, listeners made few ID change responses. However, on ID change trials listeners were nearly equal in their tendencies to make a no change and space change response. N1-P2 complexes locked to scene onsets were largest on trials where a change was indicated, but the wrong change type was chosen. P3b amplitudes locked to Scene 2 onset were monotonically related to accuracy (larger for correct and wrong change trials than change deaf trials). Effects were similar for space and ID changes, but scalp topography of P3b differed for change types (larger amplitudes for ID change trials at left temporal electrodes). Data indicate that P3b amplitude is a reliable indicator of successful change detection across different types of scene changes, showing different topography reflecting different “what” and “where” processing streams. N1-P2 complex effects suggest better encoding of scenes in relation to change detection, but unknown change type.