21st European Conference on Visual Perception
August 24-28 - 1998 - Oxford (England)
 
Ultra-rapid visual categorisation of natural scenes : No effect of familiarity on processing speed.
 
M. Fabre-Thorpe, A. Delorme, C. Marlot & S. Thorpe.
 

Centre de Recherche Cerveau et Cognition (UMR 5549) Faculté de Médecine de Rangueil, 133 route de Narbonne, 31062 Toulouse, France.

arno@cerco.ups-tlse.fr, mft@cerco.ups-tlse.fr, marlot@cerco.ups-tlse.fr, thorpe@cerco.ups-tlse.fr

 
     Monkeys and humans are highly efficient (90-94% correct) at detecting a given target (either animals or food) in a briefly flashed (20-30ms) natural image that they had never seen before. They are also strikingly fast so that the visual processing required in such a task can be done in under 150ms in humans and may be even faster in monkeys. We recently argued that such speed of processing implies that categorisation could depend mainly on fast feed-forward processing, but when a system has reached such a level of efficiency the limited number of processing stages are probably all necessary and the overall process cannot be speeded up anymore.
     Familiarity with a stimulus is one feature that can speed-up its processing. Monkeys and humans were both tested with previously unseen images that were mixed with familiar images (seen daily for three weeks for humans and even more for monkeys). they had to release a button (and touch a tactile screen for monkeys) when the image belonged to a given target category. The RT distributions for correct go responses performed towards new or familiar images were not statistically different for the two monkeys. In humans the comparison reached significance (median RT increased by 6 ms for responses towards new stimuli); but this increase was only due to targets detected with long RT. The proportion of the targets detected in less than 350ms (20%) was identical for familiar and new images and the histograms corresponding to this early responses were strictly identical. Moreover the evoked potentials showed a differential brain activity between target and non-target trials that started precisely at the same latency for new and familiar images.