A cellular mechanism for inverse effectiveness in multisensory integration

Abstract

To build a coherent view of the external world, an organism needs to integrate multiple types of sensory information from different sources, a process known as multisensory integration (MSI). Previously, we showed that the temporal dependence of MSI in the optic tectum of Xenopus laevis tadpoles is mediated by the network dynamics of the recruitment of local inhibition by sensory input (Felch et al., 2016). This was one of the first cellular-level mechanisms described for MSI. Here, we expand this cellular level view of MSI by focusing on the principle of inverse effectiveness, another central feature of MSI stating that the amount of multisensory enhancement observed inversely depends on the size of unisensory responses. We show that non-linear summation of crossmodal synaptic responses, mediated by NMDA-type glutamate receptor (NMDARs) activation, form the cellular basis for inverse effectiveness, both at the cellular and behavioral levels.DOI: http://dx.doi.org/10.7554/eLife.25392.001

Intrinsic temporal tuning of neurons in the optic tectum is shaped by multisensory experience

“…From previous studies, we knew that in tadpoles exposed to LED flashes, tectal neurons became more excitable (Aizenman et al, 2003;Ciarleglio et al, 2015). However, the stimuli we used in the present study were weaker, and similar to those used in behavioral experiments (Khakhalin et al, 2014;James et al, 2015;Truszkowski et al, 2017). We presented a checkerboard pattern that inverted once a second, for four hours; either instantaneously (dubbed "Flash"; Figure 1C left), or with a slow transition over the course of a second (old black squares shrank to white, while new black squares grew from old white squares, dubbed "Looming"; Figure 1C right).…”

“…We found ( Fig 1D, E) that exposure to these sounds (group "Sound") did not lead to significant changes in either average number of spikes (0.8±0.7 spikes; F(1,689)=1.7; p=0.2; n=28, 30), amplitude transfer function (F(1,689)=2.0; p=0.2), or temporal tuning curve (F(1,689)=2.0; p=0.2). This may suggest that acoustic stimuli did not activate tectal circuits strongly enough during conditioning, despite being more behaviorally salient (at the onset of stimulation, acoustic clicks evoked startle responses in about 50-80% of cases, compared to 5-10% for checkerboard inversions (James et al, 2015;Truszkowski et al, 2017)). This was not necessarily surprising, as mechanosensory and visual inputs have different cellular targets in the tectum Felch et al, 2016;Truszkowski et al, 2017), potentially leading to different recruitment of tectal excitatory and inhibitory circuits.…”

“…This may suggest that acoustic stimuli did not activate tectal circuits strongly enough during conditioning, despite being more behaviorally salient (at the onset of stimulation, acoustic clicks evoked startle responses in about 50-80% of cases, compared to 5-10% for checkerboard inversions (James et al, 2015;Truszkowski et al, 2017)). This was not necessarily surprising, as mechanosensory and visual inputs have different cellular targets in the tectum Felch et al, 2016;Truszkowski et al, 2017), potentially leading to different recruitment of tectal excitatory and inhibitory circuits. Acoustic stimuli may also be inherently weaker than visual stimuli in triggering plasticity effects in tectal neurons, as they arrive at different compartments within the dendritic tree (Hiramoto and Cline, 2009;Deeg et al, 2009), which may define their influence on neuronal plasticity (Richards and Lillicrap, 2019).…”

“…In the beginning of each experiment, a tadpole was put in a Petri dish (diameter of 10 cm) filled with 1-1.2 cm of tadpole rearing medium, placed on top of a CRT monitor with two speakers connected to the Petri dish with short wooden struts (James et al, 2015;Truszkowski et al, 2017), and kept there for 4 hours. The tadpole was visually isolated from…”

Experience Creates the Multisensory Transform in the Superior Colliculus

“…It is well-established that more robust unisensory responses in the SC are associated with smaller proportionate multisensory enhancements (Meredith and Stein, 1986b;Stein et al, 2009;Ohshiro et al, 2011;Otto et al, 2013;Truszkowski et al, 2017). Thus, it was possible that low ME in the dark-reared group could reflect more robust unisensory responses.…”