Electrophysiological evidence in macaque monkeys indicates that when the monkey views a visual scene with objects present in both visual hemifields, the cells of the temporal lobe respond to objects in the contralateral field, but are hardly affected by objects in the ipsilateral field. If visual memories are stored in the temporal lobes, as is generally believed, then this implies that the transfer of visual object memories from one hemifield to the other should either fail or at least suffer decrement. Building on a previous study in human subjects, we tested this prediction in rhesus monkeys

(Macaca mulaDa). We developed a method for tracking the eye movements of the awake, behaving monkey, which does not require the monkey to be restrained or surgically prepared. We optimised the system to provide reliable feedback of eye position in real time, and so provide hemifield-specific presentation of visual objects.

In each acquisition phase the monkeys learned several object discriminations concurrently, each object only ever being presented to one hemifield, and with an object present in each hemifield on every trial. In subsequent transfer tests with

the same objects, the monkeys performed significantly worse when the objects were shifted to the opposite hemifield than if shifted the same distance within one hemifield. Thus, in monkeys as well as in humans, and in association learning as well as in recognition memory, visual memories can be to a large extent hemifield-specific. This result shows that, like perceptual systems, mnemonic systems of the temporal lobe are largely hemifield-specific, and this has clear implications for studies of the temporal lobes. Further, the validation of our method will allow us to use it, in future experiments, to investigate in monkeys the effects of specific unilateral lesions on visual perception and memory for objects that are presented in known positions in the visual field.