Monash

Discovery

Visual processing relies on a balance between selectivity and invariance

25.05.2020

In a nutshell: In the earliest stages of visual processing, the brain detects and processes specific visual features by recognising a simple set of patterns.

View Paper Abstract

Visual processing relies on a balance between selectivity and invariance

The human brain has the remarkable ability to recognise specific objects, even when those objects change in appearance. For example, we can tell that a hand is a hand regardless of its colour, size, location or orientation.

When processing visual information, brain cells respond to specific features that are important to an object’s identity – that is, they display feature selectivity. At the same time, the cells ignore features that are not important – they are invariant to feature manipulation. Combining selectivity and invariance is crucial for visual processing, but how the brain does this was not well understood.

To answer this question, a team of Brain Function CoE researchers, led by Ali Almasi from the National Vision Research Institute of Australia and Hamish Meffin from the University of Melbourne, studied cells in the primary visual cortex (V1). This region of the brain is responsible for the first stage of visual processing in the cortex.

The researchers measured how the activity of cells in the V1 changed when the cells received visual information about ‘white noise’ – random combinations of black and white pixels arranged in a square grid.

Because the images of white noise are random, patterns can emerge in the pixels – such as horizontal or vertical stripes. The researchers used the brain activity data to map how the cells responded to different combinations of patterns.

The researchers built a computer model to estimate the cells’ selectivity and invariance to particular features of the different patterns, such as their orientation, frequency and phase. For a striped pattern, these features would describe whether the stripes were horizontal or vertical, how tightly spaced they were, and whether the pattern started with a black stripe or a white one.

The model revealed that most cells had a high degree of selectivity and a low degree of invariance for both orientation and frequency. However, the cells varied in their response to phase – some cells were highly selective, whereas others were completely invariant.

These findings show that even at a stage of visual processing as early as V1, the brain forms an elaborate set of sensitivities to generic features. These form the basis of more sophisticated processing in other visual areas of the brain.

Next steps:
The researchers plan to study how these selective and invariant properties develop in other regions of the brain.


Reference:
Almasi, A., Meffin, H., Cloherty, S.L., Wong, Y., Yunzab, M., & Ibbotson, M. R. (2020). Mechanisms of feature selectivity and invariance in primary visual cortex. Cerebral Cortex, bhaa102. doi: 10.1093/cercor/bhaa102


Republish this article:

We believe in sharing knowledge. We use a Creative Commons Attribution 4.0 International License, which allows unrestricted use of this content, subject only to appropriate attribution. So please use this article as is, or edit it to fit your purposes. Referrals, mentions and links are appreciated.

CIBF