For both considered models of Figure 1, the response rises with increasing
overall stimulus intensity in a sigmoid fashion (Figure 1C), as determined by the intrinsic nonlinearitities of each output neuron. Despite these similarities in the general shape of the stimulus-response relations, the characteristic differences of stimulus integration in the two models become strikingly apparent when one considers the contour lines of the stimulus-response plot, that is, the lines along which the response of the output neuron stays the same (Figures 1A and 1B, shown below the stimulus-response surface plots). The shape of these iso-response curves is a clear signature of the underlying signal integration or, in other words, of the arithmetic Selleckchem Ixazomib rule with which the output neuron combines its inputs.
In the simplest case, linear summation of inputs is reflected HA-1077 by straight lines in the iso-response curves (Figure 1A). The circular part of the iso-response curves in Figure 1B, on the other hand, shows the summation of squared positive inputs, whereas the line segments that run parallel to the axes indicate the thresholding of negative inputs. Iso-response curves thus reveal the nature of stimulus integration independently of the neuron’s intrinsic output nonlinearity; the output nonlinearity simply affects the response equally for all stimuli along an iso-response curve and thus does not influence the curve’s shape. To assess the nature of signal integration within the receptive field center of retinal ganglion cells, we developed an approach to measure these iso-response curves. We used a stimulus layout that subdivided the receptive field center of a ganglion cell into two halves and stimulated the cell with different levels of light intensity in these two regions (Figure 1D). Iso-response curves then consisted Thiamine-diphosphate kinase of those pairs of
visual contrast in the two receptive field halves (measured relative to the mean background light intensity) that yielded a fixed, predefined spike response of the ganglion cell. Seeking iso-response stimuli poses an obvious experimental challenge; instead of measuring responses for predefined stimuli, we need to find stimuli for predefined neuronal responses. To achieve this, we devised a closed-loop experimental design to automatically and quickly tune stimulus intensities toward the desired response, similar to previous applications in the auditory system (Gollisch et al., 2002 and Gollisch and Herz, 2005). We recorded spiking activity extracellularly from individual ganglion cells in isolated salamander retinas. For every analyzed cell, we first used the online analysis to map out the location and size of the cell’s receptive field center.