Emergence of Selectivity to Looming Stimuli in a Spiking Network Model of the Optic Tectum

Abstract

The neural circuits in the optic tectum of Xenopus tadpoles are selectively responsive to looming visual stimuli that resemble objects approaching the animal at a collision trajectory. This selectivity is required for adaptive collision avoidance behavior in this species, but its underlying mechanisms are not known. In particular, it is still unclear how the balance between the recurrent spontaneous network activity and the newly arriving sensory flow is set in this structure, and to what degree this balance is important for collision detection. Also, despite the clear indication for the presence of strong recurrent excitation and spontaneous activity, the exact topology of recurrent feedback circuits in the tectum remains elusive. In this study we take advantage of recently published detailed cell-level data from tadpole tectum to build an informed computational model of it, and investigate whether dynamic activation in excitatory recurrent retinotopic networks may on its own underlie collision detection. We consider several possible recurrent connectivity configurations and compare their performance for collision detection under different levels of spontaneous neural activity. We show that even in the absence of inhibition, a retinotopic network of quickly inactivating spiking neurons is naturally selective for looming stimuli, but this selectivity is not robust to neuronal noise, and is sensitive to the balance between direct and recurrent inputs. We also describe how homeostatic modulation of intrinsic properties of individual tectal cells can change selectivity thresholds in this network, and qualitatively verify our predictions in a behavioral experiment in freely swimming tadpoles.

Graph analysis of looming-selective networks in the tectum, and its replication in a simple computational model

“…There was no discernible difference in response amplitudes between looming and scrambled stimuli (average difference of −0.03±0.15 and −0.01±0.20; p t1 = 0.45 and 0.78 for younger and older animals respectively; no difference in development p t = 0.80). These results support our prior observation that total tectal responses in tadpoles depend mostly on the dynamics of visual stimuli, rather than on their geometry (Khakhalin et al, 2014;Jang et al, 2016).…”

“…When looming stimuli were presented to the model, they "resonated" with matching synfire chains, causing them to respond strongly. It may be that in the biological tectum, enhanced responses to looming stimuli are due to either delayed recurrent integration (Khakhalin et al, 2014;Jang et al, 2016), dynamic inactivation of neurons (Fotowat et al, 2011), or some other non-linear effects (Baginskas and Kuras, 2009). Two biggest discrepancies between the model and the experiments were the position of selective cells within the network (central in tadpoles, peripheral in the model; Figure 3C vs. Figure 8J), and the difference in centrality measurements related to local signal integration (slightly higher in-degree and Katz rank in biological experiments, but no similar effect in the base model; Figure 6A-C vs. Figure 8N,O).…”

“…Contrary to our expectations, and in contrast to what is known about visual inputs to the tectum (Tao and Poo, 2005), the intra-tectal connectivity did not become more compact in development (p t =0.7). This suggests that tectal networks rely on relatively far-reaching recurrent connections to integrate visual information across the visual field (Baginskas and Kuras, 2009;Liu et al, 2016;Jang et al, 2016). A.…”

“…Across species, looming detection relies on a diverse set of mechanisms that include dimming detectors (Ishikane et al, 2005;Münch et al, 2009), integration of opponent motion (Klapoetke et al, 2017), and competitive spike-frequency adaptation (Peron and Gabbiani, 2009;Fotowat et al, 2011). Even within a single clade of anuran amphibians (frogs), animals seem to employ at least two competing approaches to loom detection: a non-linear response to dimming-induced retinal oscillations (Baranauskas et al, 2012), and a rebound of recurrent activity during edge expansion (Khakhalin et al, 2014;Jang et al, 2016). Moreover, at least in some cases, competing mechanisms may lead to different motor responses, as described in insects (Card and Dickinson, 2008;Chan and Gabbiani, 2013), fish (Budick and O'Malley, 2000;Burgess and Granato, 2007;Portugues and Engert, 2009;Temizer et al, 2015;Bhattacharyya et al, 2017), and tadpoles (Khakhalin et al, 2014).…”

“…First, we can look for synfire chains in the tectum; see whether they get selectively activated in response to looming stimuli, and check whether their connectivity and selectivity patterns change with development. Luckily, signal transmission in Xenopus tectal neurons is relatively slow, with synaptic transmission and spiking initiation taking up to 10 ms (Ciarleglio et al, 2015;Jang et al, 2016;Busch and Khakhalin, 2019), so if synfire chains in the tectum exist, they may be directly observable with fast calcium imaging, operating at rates of ∼100 frames/s. Second, if synfire chains can serve as a basis for looming detection, we can expect looming selectivity to spontaneously emerge in a model of a developing tectum governed by spike-time-dependent plasticity (Gao and Ganguli, 2015;Pietri et al, 2017).…”