As described in the prior sections, varying the last pulse affects OSR significantly, and the delay between probe and entrainment also varies the latency[77]. These re-sults reflect the sophisticate encoding for temporal patterns in the retina. We further test the response by presenting complex temporal patterns with multiple periods.
The synchronization process and OSR-like response after terminating the stimuli are compared. Furthermore, some preliminary activities under spatial patterns are reported.
3.3.1 Heterogeneous response to multiple periods
Stimuli with multiple periods are constructed by repeating certain intervals and fixing the sequence. For instance, three different inter-pulse-intervals (220, 180, and 140 ms indicated by symbols A, B, and C) were presented periodically as an entrainment (repeating the order of ABC ten times). Under such stimuli, we find that the retina is able to recognize, adapt to, and possibly predict certain complex input patterns. The firing rate to each reoccurring interval changes in the first 4 seconds, reaching a steady state after a time scale of seconds. If the period is equal to A+B+C intervals, the post-stimuli responses is less significant. Also, the same number of pulses and intervals where randomly permutated for comparison.
The entrained process induces higher firing rates than entrainment with random permutated intervals (Fig. 3.16).
While the activity under these complex temporal patterns present rich dynam-ics, the timing of the post-stimulation firing peak is similar to OSR after periodic stimuli. The latency is still mainly reflecting the last pulse. However, it is shown
Figure 3.16: PSTH for complex temporal patterns. (a) Periodic pulses with pe-riod=180 ms. (b) “ABC” repetitive temporal pattern, where A=220 ms, B=180, C=140 ms. (c) Periodic stimuli with “A+B+C” period=540 ms. (d) Randomized
“ABC” pattern with the same number of pulses and intervals shown in (b). Averaged activities are calculated from 59 recorded units.
even though the last interval is fixed and the latency to OSR-like activity is similar, the post-stimulation activities over channels show significant higher heterogeneity compare to the highly synchronized OSR after periodic stimulation (Fig. 3.17).
3.3.2 Effects of spatial patterns
By measuring the cross-correlation between spike trains recorded from different channels, the conclusion from the last session can be visualized. Certain clusters of channels synchronize higher only when the stimuli contains multiple periods. It is possible that certain temporal input can be encoded into a population of cells in the retina. Is it true in another way around, if spatial patterns affect the entrainment of temporal patterns? We test this by replacing the spatially homogeneous LED light flash into a LCD that projects spatial patterns onto the retina. Four kinds of spatial patterns were presented as a periodic temporal input: spatially uniform fields (as a
Figure 3.17: Heterogeneous response to complex temporal patterns. (a) Heat map for firing rates averaged over 20 trials from 60 channels. Horizontal axis starts from the timing of the last pulse. The upper plot is for ABC stimuli (A=220, B=180, C=140 ms) and lower plot is for periodic stimuli with period=180 ms (could be seen as “BBB”). Inset in each plot is the cross-correlogram for 60 channels calculated from the post-stimuli spike trains, with reddish color codes showing higher correlation coefficients. (b) Average PSTH for the same sample shown in (a) that focuses on timing after the last pulse. Post-stimuli firing patterns for four stimulations plotted in Fig. 3.16 are shown.
control), checkerboards, anti-checkerboards, and random mosaics.
Specifically, these patterns are presented for 50 ms following a 50 ms black whole field image (period=100 ms) and repeated for 20 time, just as how OSR experiments are done with LED stimulation. For checkerboards, the black and white checkers are 60 µm. The anti-checkerboard stimulation presents two checkerboards that are anti-phase, meaning the each unit in these checkerboard are flashing in the same period (100 ms) but with opposite phase from the 8 neighbors. For the random mosaic stimuli, each frame randomly presents white and black unites in the same checkerboard coordinate, maintaining the average intensity but destroying the reg-ular spatial information. These designed stimulation are used to identify the func-tional spatial integration for temporal entrainment. The possible out comes to be tested are: OSR is produced by local units (within a receptive field of the retinal ganglion cell) so would occur when the checkerboard is used for entrainment, or in contrast, OSR is a a collective phenomenon that requires the whole patch of retinal sample to be synchronized and would be disrupted by the spatial heterogeneity.
OSR still occurs when the periodic stimuli contains spatial patterns such as the checkerboard. It is shown that the timing and synchrony are not affected, similar to the control group using spatially uniform stimulation. However, anti-checkerboard that provide neighbor pixels oscillating out of phase cancels OSR. There is little synchronized post-stimulation activities after such stimulation. On the other hand, neither can randomly organized pixels provide robust entrainment for OSR. It is noteworthy that the random patterns within the same trial are different for all flashes. If we fix the pattern within a trial but only change the pattern between trials, the PSTH still shows OSR-like synchronized firing patterns after the stimuli (Fig. 3.18).
Figure 3.18: Periodic stimuli with spatial patterns. (a) Whole field periodic stimuli.
(b) Periodic stimuli that flashes a checkerboard pattern. (c) Periodic stimuli with each consecutive pattern showing an anti-phase checkerboard pattern. (d) Periodic stimuli that flashes random patterns with 50% white pixels randomly distributed.
All the stimuli have period=100 ms and the average firing rate PSTH are calculated from 59 recorded units.