Synthesizing perception of 3-D motion from 2-D sense input – lessons from the electrosensory system


October 10, 2017 - 12:00pm
Northwest Building, Room 243
About the Speaker
Leonard Maler
Speaker Title: 
Speaker Affiliation: 
University of Ottawa

Electric fish are able to locate objects undergoing 3-D motion by computing the 2-D ‘electric images’ that such objects project onto their body surface. Electrolocation utilizes an electric organ and electroreceptors. The electric organ discharge (EOD) generates an electric field around the fish’s body that is sensed by electroreceptors distributed over its body surface (~7500/side). A conductive object (e.g., prey, positive contrast) locally increases transdermal electric current flow across a small patch of skin. This is a ‘bright electric image’ and it evokes an increase in the firing rate of electroreceptor afferents (EAs) in that skin patch. A non-conductive object (e.g., a rock, negative contrast) does the opposite – producing a ‘dark electric image’ that decreases the EA firing rate. Electrosensory circuitry transforms the 2-D spatio-temporal pattern of EA discharge to a representation of 3-D motion that directs the fish’s navigation and prey capture movements. EAs project to pyramidal cells within multiple topographic maps in the hind brain electrosensory lobe (ELL); pyramidal cells also receive strong feedback input. Pyramidal neurons come in two flavours: ON cells respond strongly to motion of ‘bright’ objects (brass spheres) while OFF cells respond strongly to motion of ‘dark’ objects (plastic spheres); the putative labelled line encoding of positive (ON cell) versus negative (OFF cell) contrast seen in the retina is therefore also hardwired into ELL circuitry. Remarkably, upon reversal of object motion, e.g., from looming to receding, there is a dramatic coding switch: ON cells respond to receding motion of plastic spheres (negative contrast) while OFF cells respond to receding motion of brass spheres (positive contrast). Despite the breakdown in labelled line coding, the responses to looming and receding motion are mirror symmetric and the distance of an object from the fish’s skin can be accurately decoded from the ON and OFF cell firing rates. Upon motion reversal, the EA response becomes distorted. The accurate response to receding motion is merely triggered by EA input, but then completely generated by a time limited positive feedback loop emanating from the midbrain.

Conclusion: Electric fish accurately estimate the location and velocity of, e.g., moving prey. The electrosensory input is ambiguous because 3-D information is lost upon projection onto a 2-D sensory surface. The veridical estimates of location and velocity are initiated by sensory input but also require active 3-D reconstruction by the intrinsic dynamics of positive feedback loops.