The responses of neurons in the cortex in the superior temporal sulcus of the monkey to band-pass spatial frequency filtered faces

Edmund T. Rolls,Gordon C. Baylis,Michael E. Hasselmo
DOI: https://doi.org/10.1016/0042-6989(87)90081-2
IF: 1.8
1987-01-01
Vision Research
Abstract:There are neurons in the cortex in the anterior part of the superior temporal sulcus of the macaque monkey with visual responses which would be useful for face recognition (Rolls, 1984; Baylis et al. , 1985). To analyze further the information which leads them to respond, their responses were measured to parametrically filtered stimuli. The responses of 48 such single neurons were measured to faces which were digitized and were bandpass spatial frequency filtered. The octave width bands were 2–4, 4–8, 8–16, 16–32, 32–64 and 64–128 cycles per image. It was found that the neurons could respond well to single octaves of the spatial frequencies normally present in faces, that the most effective bands were 4–8, 8–16 and 16–32 cycles per face (cpf), and that the bands 2–4 and 32–64 cpf were partly effective. In investigations of whether the responses of the neurons to an unfiltered face, and to low-pass and high-pass filtered images could be predicted by linear addition of their responses to each of the octave bands shown separately, it was found that the majority of the neurons were non-linear, and responded much less than predicted. It was also shown that this occurred even when the contrast was reduced to 0.25 of that normally present in a face, so that the result was not due just to a ceiling effect of the maximum firing rate. These results help to define parametrically the aspects of the information normally present in a face which are sufficient to produce responses of these neurons to them, and show that linear operations cannot account for information processing in this part of the visual system.
What problem does this paper attempt to address?