Qualitative similarities and differences between monkey and human visual perception

Abstract

Animal models play a crucial role in helping neuroscientists unravel the computations supporting behaviour at the resolution of a single neuron in the brain. In vision science, the macaque monkey is the pre-eminent model organism not just due to the similarities in the brains structure but also due to the rich behavioral repertoire display by them. A rich history of neurophysiological experiments have revealed the complex computations supporting vision from the thalamus all the way to the pre-frontal cortex. Yet, we do not know enough about how monkeys perceive the visual world. Increasingly, contemporary research has highlighted differences between monkey and human perception. While these observations have highlighted the need to carefully examine comparisons across the two species, it is not clear how the varying task paradigms (employed in these studies) affect the conclusions.

In my thesis, I performed a series of studies to understand the similarities and differences between monkey and human visual perception. In the first study, I trained monkeys on an oddball visual search task - a very intuitive easy to learn task. Humans are also adept at this same task. With a common task paradigm in hand, I then tested monkeys and humans across a range of perceptual phenomena - from general to specific. I observed that while monkeys shared general perceptual properties of vision with humans - like Weber’s Law, animate-inanimate organization and amodal completion, they differed from humans in how they perceived many special phenomena. Monkeys did not integrate local parts into a whole nor did they show a clear effect of mirror confusion. This indicates that monkeys and humans share similar perception in general but also diverge significantly with important consequences.

In the second study, I investigated the fidelity of amodal completion in humans. Amodal completion is an omnipresent phenomenon which supports our perception of occluded objects - by representing the occluded portions in a sensible manner. In humans, amodal completion has been demonstrated for simple edges and shape contours, but we did not know how complex features are represented behind the occluder. Using a novel within display conditions paradigm in visual search, I observed that humans indeed complete complex contours behind the occluder. This completion is not driven by symmetry cues and is dependent of the visible features of the occluded shape.

Finally, in the third study, I examined amodal completion in monkeys. Here, I trained monkeys on a novel free-choice paradigm to report what they see behind the occluded shape. I took great care to reduce the effects of any task-related bias, training bias in monkey’s reports of what they see behind the occluder. Like human’s monkeys also perceived complex features behind the occluder which was dependent on the visible features on the occluded display. Furthermore, we showed through another free-choice task that humans also chose similar completions for occluded objects like monkeys.

To summarize, these studies show that while there is a general agreement between human and monkey visual perception, there are instances where they diverge and in telling ways. Importantly, as monkeys are widely used to understand vision at the neural level, understanding how and where such differences arise in the brain is an important stepping stone in the quest to answer how human brains process the visual world.