| Abstract: | Through a series of original research articles, this thesis describes the role of neural
network feedback in top-down control of bursty ON and OFF type electrosensory
neurons while processing motion toward (looming) and away (receding) from the
body. Neural codes for motion reversal in weakly electric fish are not simply evoked
by bottom-up sensory input and amplified by feedback; instead, positive feedback
loops must synthesize a directionally invariant representation of motion reversal that
is distributed across both the ON and OFF pyramidal cell populations. Through
balanced excitatory and inhibitory feedback, the system establishes an optimal distance for motion estimation (a sensory focus) that is maintained by the animal during a motion tracking behaviour. Remarkably, this sensory focus is size, direction and speed-invariant. The speed invariance likely derives from the speed invariance of
the electrosensory afferent response, a consequence of timescale-free spike frequency adaptation. Since natural swimming movements are associated with tail-bending that cause spatially diffuse sensory noise, we demonstrate that spatially localized motion processing by the ON and OFF neurons co-occurs with cancellation of the distraction; this supports the circuit's role as a robust `sensory searchlight' mechanism for spatial attention. A simple algorithm for motion tracking is discussed, as well as potential generalizations of the described coding principles to more complex mammalian circuits. |