MAGNITUDE RATINGS FOR PERCEIVED CHANGES IN PHOTOGRAPHS OF NATURAL SCENES MAY BE LINEARLY PROPORTIONAL TO DIFFERENCES IN NEURONAL FIRING RATES
AbstractWe are studying how people perceive suprathreshold changes in the colour, size, shape orlocation of items in images of natural scenes. We use magnitude estimation ratings tocharacterise the sizes of the perceived changes in thousands of photographs, and we havebuilt a computational model that tries to explain observersâ€™ ratings of naturalistic differencesbetween image pairs. We model the action-potential firing rates of millions of neurons,having linear and non-linear summation behaviour closely modelled on real visual-cortexneurons. Although the model is still imperfect, it does produce tolerable predictions of theratings for most kinds of image change. Importantly, ratings rise roughly linearly with themodelâ€™s numerical output, which represents differences in neuronal firing rate in response tothe two images under comparison. While rating may not be directly proportional to metrics ofstimulus difference, it seems to be proportional to the neuronal response.