Mirror-Image Confusion in Single Neurons of the Macaque Inferotemporal Cortex

Abstract

Humans and animals confuse lateral mirror images, such as the letters "b" and "d," more often than vertical mirror images, such as the letters "b" and "p." Experiments were performed to find a neural correlate of this phenomenon. Visually responsive pattern-selective neurons in the inferotemporal cortex of macaque monkeys responded more similarly to members of a lateral mirror-image pair than to members of a vertical mirror-image pair. The phenomenon developed within 20 milliseconds of the onset of the visual response and persisted to its end. It occurred during presentation of stimuli both at the fovea and in the periphery.

Breaking the symmetry: Mirror discrimination for single letters but not for pictures in the Visual Word Form Area

“…There is evidence that left-right mirror invariance is deeply entrenched in the ventral visual pathway of all primates, as inferior temporal cortex neurons in monkeys show such an invariance even when using abstract geometric forms as stimuli: neurons that respond preferentially to a given picture also respond in an nearly-equivalent manner to their mirror counterpart, but not to other viewpoints (Logothetis et al, 1995;Rollenhagen and Olson, 2000;Baylis and Driver, 2001). Moreover, mirror invariance was also demonstrated in infants (Bornstein et al, 1978) and may therefore constitute a default invariance of the visual system in humans.…”

“…Inter-hemispheric connections through the corpus callosum, linking symmetrically regions of each hemisphere point by point, would operate in such a way that a "b" in one hemisphere would become a "d" in the other (Corballis and Beale, 1976;Dehaene, 2009). Others authors (Rollenhagen and Olson, 2000;Davidoff and Warrington, 2001) stress that mirror discrimination, for identification purposes, does not represent any evident advantage in a natural world when the vast majority of living and non-living forms do not change category under mirror symmetry. It is therefore plausible that the orientation information is simply not coded at all, so that separate representations of original and mirror representations do not exist, which seems coherent with single neuron monkey data (Logothetis et al, 1995;Rollenhagen and Olson, 2000).…”

“…Others authors (Rollenhagen and Olson, 2000;Davidoff and Warrington, 2001) stress that mirror discrimination, for identification purposes, does not represent any evident advantage in a natural world when the vast majority of living and non-living forms do not change category under mirror symmetry. It is therefore plausible that the orientation information is simply not coded at all, so that separate representations of original and mirror representations do not exist, which seems coherent with single neuron monkey data (Logothetis et al, 1995;Rollenhagen and Olson, 2000). For letters, however, such distinct representations would have been created, for instance based on the necessity to map them onto distinct phonological or gesturebased codes in higher areas.…”

“…Humans, monkeys, pigeons and even octopuses can easily recognize reflected mirror images of pictures previously exposed, although these mirror images induce a very different retinal projection (Sutherland, 1960;Mello, 1965;Hollard and Delius, 1982;Tarr and Pinker, 1989;Logothetis et al, 1995;Rollenhagen and Olson, 2000;Baylis and Driver, 2001;Fiser and Biederman, 2001). This spontaneous "mirror invariance" phenomenon enables one to recognize images which are indifferently seen from a left or right perspective (e.g.…”

“…The brain's underpinnings of mirror invariance are starting to be uncovered in non-human primates and humans. Monkey's single cell recordings show neurons in visual ventral cortex (IT) responding in an equivalent way to a given picture and its mirror reflected presentation (Logothetis et al, 1995;Rollenhagen and Olson, 2000). In humans, the phenomenon of fMRI repetition suppression has been used to demonstrate that pictures and their mirror images can be encoded as the same object within the ventral stream (Eger, 2004;Vuilleumier et al, 2005;Dehaene et al, 2010).…”