Burst Firing in a Motion-Sensitive Neural Pathway Correlates with Expansion Properties of Looming Objects that Evoke Avoidance Behaviors

Abstract

The locust visual system contains a well-defined motion-sensitive pathway that transfers visual input to motor centers involved in predator evasion and collision avoidance. One interneuron in this pathway, the descending contralateral movement detector (DCMD), is typically described as using rate coding; edge expansion of approaching objects causes an increased rate of neuronal firing that peaks after a certain retinal threshold angle is exceeded. However, evidence of intrinsic DCMD bursting properties combined with observable oscillations in mean firing rates and tight clustering of spikes in raw traces, suggest that bursting may be important for motion detection. Sensory neuron bursting provides important timing information about dynamic stimuli in many model systems, yet no studies have rigorously investigated if bursting occurs in the locust DCMD during object approach. We presented repetitions of 30 looming stimuli known to generate behavioral responses to each of 20 locusts in order to identify and quantify putative bursting activity in the DCMD. Overall, we found a bimodal distribution of inter-spike intervals (ISI) with peaks of more frequent and shorter ISIs occurring from 1-8 ms and longer less frequent ISIs occurring from 40-50 ms. Subsequent analysis identified bursts and isolated single spikes from the responses. Bursting frequency increased in the latter phase of an approach and peaked at the time of collision, while isolated spiking was predominant during the beginning of stimulus approach. We also found that the majority of inter-burst intervals (IBIs) occurred at 40-50 ms (or 20-25 bursts/s). Bursting also occurred across varied stimulus parameters and suggests that burst timing may be a key component of looming detection. Our findings suggest that the DCMD uses two modes of coding to transmit information about looming stimuli and that these modes change dynamically with a changing stimulus at a behaviorally-relevant time.

Keywords: DCMD; bursting; locust; neuron; sensory coding; vision.

Figure 1

Descending contralateral movement detector (DCMD) responses time aligned to the perceived time of collision (TOC, red dashed line) of a laterally looming visual stimulus. (A) Raw extracellular neuronal recording representing a typical DCMD response (large spikes) to our looming stimulus. Note the presence of multiple bursts (gray squares) throughout the recording. Inset—Inter-spike Intervals (ISI) histogram from a single presentation to one animal highlighting that most of the ISIs are within 8 ms (shaded area), which defines spikes within bursts (Quantification of a burst is described in methods). (B) Raster plot (n = 600 responses) with overlay of a 50 ms Gaussian smoothed (1 ms bin width) mean full DCMD firing rate (black line), burst firing rate (red line) and isolated single spike firing rate (blue line). Each row of rasters represents the response of a single DCMD neuron and each raster (in gray) represents a single spike (N = 20 animals with 30 presentations to each). Rasters were organized in ascending order based on the timing of the spike before TOC. Note the consistent vertical banding pattern in the raster plot. Also note that the more distinctive oscillations in the mean response end around 200 ms before TOC (black dashed line). Note the decline in single spike firing rate after t = 200 ms, where bursting dominates and peaks around TOC. (C) Vertical step line plot representing the change in subtense angle of the looming stimulus. As the edges of the virtual disc expanded, the full DCMD response in (B) increased to a peak that occurred before TOC.

Figure 2

Comparison of interspike intervals (ISIs) during the entire stimulus. (A) This ISI return map compares one ISI (y-axis) with the following ISI (x-axis) and displays a fan-shaped distribution of ISIs with a tight cluster of points at the origin (in red) and two other clusters along the axes (cyan and blue). Clustering along each axis was defined by an 8 ms ISI within a burst. This type of distribution is typical of bursting neurons, where a cluster along the y-axis (cyan) represents first ISIs of each burst, a cluster along the x-axis (blue) represents the ISIs that follow bursts, the cluster in the bottom left corner (red) represents intra-burst ISI and the more scattered clusters (black) represents ISIs between isolated spikes. (B) Joint ISI distribution heat map showing local density of clusters (smoothed with a 3 ms Gaussian radius) and highlighting the clusters identified in (A). The color scale to the right of the graph defines relative local densities. (C) ISIs tended to shorten as the stimulus reached TOC (red vertical line), however there was a relatively constant ISI distribution in short intervals along the x-axis (pink arrows). During the final 200 ms before TOC, the ISIs became progressively shorter, which mirrors the mean DCMD plot in Figure 1B.

Figure 3

ISI histograms (ISIh; left panels) and autocorrelations (right panels) of DCMD responses for full responses (A) DCMD bursts (B) and isolated spikes (C). Data for ISIh (counts/bin) were normalized to the number of interspike intervals within the train of events (bursts or spikes, see “Materials and Methods” Section). While partially masked within the full DCMD response in all stimulus epochs (A) there is a bimodal distribution of burst intervals (B) in the full DCMD response (black line) and in bursts during the final 200 ms before TOC (blue line); one with shorter ISIs and a second with longer ISIs around 40 ms. This trend is more clearly visualized in the DCMD’s 100-ms autocorrelation. Following identification of bursts, we found a relatively unimodal distribution with highest rates occurring around 40 ms; this trend was also reflected in the associated autocorrelations. Overall, there was no clear trend in the distribution of isolated spikes.

Figure 4

Statistical comparisons of DCMD response types (full responses, bursts, and isolated spikes). The amplitude and time of peak firing rate (A,B), total number of events (spikes or bursts), (C) and peak width at half height (PWHH); (D) were all significantly different between each of the response types. The values reported for the full DCMD response are consistent with previously published work (see “Results” Section). The isolated spike rate generally increased up to 200 ms before TOC, generating a relatively short but high peak firing rate, while bursting generally increased and peaked from 200 ms before TOC to TOC. Different letters above or below bars or boxes represent significant differences between parameters within each panel. Significance assessed at P < 0.05.

Figure 5

Statistical comparisons of burst structure. We performed a burst analysis on three phases of the DCMD response to looming: during the entire stimulus duration (white boxes) and the two stimulus epochs: up to 200 ms before TOC (light gray boxes), and from 200 ms before TOC to TOC (dark gray boxes). Examination of each burst revealed that as the stimulus approached the locust, there was an increase in burst duration (A), spikes in bursts (B), peak firing rate in bursts (D) and percentage of total spikes contained in a bursts. (F) However, the interval between bursts (E) decreased as the stimulus approached TOC and the ISI within each burst (C) was similar for all three phases. Different letters above boxes represent significant differences between parameters within each panel. Significance assessed at P < 0.05.

Figure 6

Bursting in response to varied stimulus expansion parameters. We applied our bursting algorithm to a subset of data from Dick and Gray (2014) and plotted the mean firing rate (n = 20 locusts, see “Materials and Methods” Section) of the full response (A), bursts (B), spikes in bursts (C) and isolated spikes (D) relative to the time of collision (TOC; dashed vertical line). The inset in (A) identifies the l/|v| value of each plot in each panel. Filled circles in each plot represent the time of the peak or valley.

Figure 7

Relationship between expansion parameters and firing properties. For each l/|v| value, we plotted the firing rate at the peak of the full response, bursts, spikes in bursts and isolated spikes (A, top panel) as well as the firing rate at the valley of the isolated spikes (A, bottom panel). We also plotted the time of the relative peaks (B, top panel) or valleys (B, bottom panel) against l/|v|. Data represent mean values from 20 locusts and were fit with linear regression lines (see Table 1 for regression parameters).