The remarkable pattern recognition abilities of primates (humans and non-humans) surpass the best computational algorithms available today.
The difficulty of visual recognition stems from the need to achieve high selectivity in hundreds of milliseconds while maintaining robustness to object transformations. We record neurophysiological activity from the human temporal lobe in subjects who suffer from intractable epilepsy; this provides a window of opportunity to examine activity in the human brain at high spatial and temporal resolution.
Recording activity from thousands of electrodes in tens of human subjects, we quantified at high temporal resolution the amount of information conveyed by intracranial field potentials in human visual cortex. Subjects were presented with images containing one or more objects or with movies. We could accurately decode object category information in single trials as early as ~100 ms post-stimulus. Decoding performance was robust to changes in rotation, scale and clutter. Furthermore, visual information could also be decoded under dynamic viewing conditions during movies. The results revealed that physiological activity in the human temporal lobe can account for some of the key properties of visual recognition and provide strong constraints for computational models of human vision.
We compare the neurophysiological results against simple hierarchical and feed-forward models of visual object recognition.