On cardinal directions in spatial pattern space and falsifying multi-channel detection models

It has been generally recognised that in its early stages the human visual system comprises a set of independent subsystems, or channels, acting in parallel. There is general agreement that the receptive fields of neurones constituting the channels overlap considerably, making the task of selectively stimulating individual channels nontrivial. Since such a task is important for estimating the spatio-temporal characteristics of these channels, a method for determining a set of spatial patterns which stimulate multiple channels independently, irrespective of how their receptive fields overlap, is presented here. As an example such patterns were calculated for Wilson and Bergen's model (Wilson and Bergen, 1979, Vision Res. 19, 19-32). Using the modification of the subthreshold summation technique (Logvinenko, 1995, Biol. Cybernet. 73, 547-552) it is shown that in reality these stimuli are not processed independently. It follows that Wilson and Bergen's model involves a set of channels which is inappropriate or incomplete or both.