The Results

Recognising faces or just photos?

We have recently completed an experiment where we compared same and different photos of unfamiliar faces in an item recognition task. We predicted that memory for different faces would have to be supported by recollection of facial information whereas memory for same photos could be supported by recollection of facial information, familiarity for pictorial details of the stimulus, or a combination of the two sources of mnemonic information.

The ability to discriminate between old and new stimuli is better for same photos than for different photos. We believe that the superior performance for same photos is due to the availability of stimulus-driven, perceptual familiarity.

The ability to discriminate between old and new stimuli is better for same photos than for different photos. We believe that the superior performance for same photos is due to the availability of stimulus-driven, perceptual familiarity.

Participants studied 200 unfamiliar faces and made old/new decisions to 100 same, 100 different and 200 new faces across 10 blocks. While 70% of the same photos were recognised, only 54% of the different photos were correctly recognised. The superior performance for the same photos can be seen in Discriminability figure below, which corrects the raw hit rates for false alarms.

The superior performance for same photos was also associated differences in ERP old/new effects between conditions, centring on the modulation of old/new effects over posterior scalp electrodes from 500-500msec after stimulus-onset. Both same and different photos produced ERP activity over anterior sites from 300-500msec. In the subsequent 500-700msec latency period, activity at the front of the head is the same for both conditions, but at the back of the head there is a greater amount of neural activity for same photos than for different photos (see topographic maps).

The difference in voltage between old and new waveforms is plotted separately for same and different photos. Each scalp map reflects a view of the head as if looking down form behind (top = front, left = left). The scale bar represents the size of the old/new difference in microvolts, with dark red indicating larger effects. In the 300-500msec latency period, neural activity associated with recognising same and different photos is clearly pronounced over frontal scalp sites, whereas in the 500-700msec latency period the activity for same photos extends more the the back of the head than the activity for different photos.

The difference in voltage between old and new waveforms is plotted separately for same and different photos. Each scalp map reflects a view of the head as if looking down form behind (top = front, left = left). The scale bar represents the size of the old/new difference in microvolts, with dark red indicating larger effects. In the 300-500msec latency period, neural activity associated with recognising same and different photos is clearly pronounced over frontal scalp sites, whereas in the 500-700msec latency period the activity for same photos extends more the the back of the head than the activity for different photos.

 

The difference in voltage between old and new waveforms is plotted separately for same and different photos. Each scalp map reflects a view of the head as if looking down form behind (top = front, left = left). The scale bar represents the size of the old/new difference in microvolts, with dark red indicating larger effects. In the 300-500msec latency period, neural activity associated with recognising same and different photos is clearly pronounced over frontal scalp sites, whereas in the 500-700msec latency period the activity for same photos extends more the the back of the head than the activity for different photos.

Based on previous literature reported by MacKenzie and Donaldson (2007; 2009), we interpret the anterior portion of the old/new effect as evidence of recollection contributing to task performance, whereas the posterior activity reflects the presence of familiarity for same photos only. On the basis of these data we argue that the reason why recognition memory for unfamiliar faces is relatively difficult is that there is no familiarity information available and people are forced to rely on recollection. Recognition memory for photos of unfamiliar people tested with the same photos, by contrast, is relatively easier because both familiarity and recollection can provide sources of mnemonic information.

Why do we see an anterior ERP old/new effect for unfamiliar faces rather than the classic left parietal effect? Galli and Otten (2011) observed an anterior effect for unfamiliar faces and for pictures of common objects, and therefore we believe that the anterior effect reflects recollection of visual information that is not consolidated into long-term memory.