On the organization of receptive fields of retinal spot detectors projecting to the fish tectum: Analogies with the local edge detectors in frogs and mammals

Abstract

Responses of ON‐ and OFF‐ganglion cells (GCs) were recorded extracellularly from their axon terminals in the medial sublamina of tectal retino‐recipient layer of immobilized cyprinid fish (goldfish and carp). These units were recorded deeper than direction selective (DS) ones and at the same depth where responses of orientation selective (OS) GCs were recorded. Prominent responses of these units are evoked by small contrast spots flickering within or moving across their visual field. They are not selective either to the direction of motion or to the orientation of stimuli and are not characterized by any spontaneous spike activity. We refer to these fish GCs as spot detectors (SDs) by analogy with the frog SD. Receptive fields (RFs) of SDs are organized concentrically: the excitatory center (about 4.5°) is surrounded by opponent periphery. Study of interactions in the RF has shown that inhibitory influences are generated already inside the central RF area. This fact suggests that RFs of SDs cannot be defined as homogeneous sensory zone driven by a linear mechanism of response generation. Physiological properties of fish SDs are compared with the properties of frog SDs and analogous mammalian retinal GCs—local edge detectors (LEDs). The potential role of the SDs in visually guided fish behavior is discussed.

Direction selectivity of the retinotectal system of fish: Findings based on microelectrode extracellular recordings of the tectum opticum

“…With the new setup, we managed to accurately determine the depth at which we could record different retinal GC projections in the tectal retinorecipient layer. Thirteen types of GCs projecting to the fish TO were described [1,16,23,[39][40][41][42] (Fig. 10B).…”

“…The deepest sublayer (at a depth of about 200 μm): Two types of sustained units are shown -those activated by the darkening(12) and others activated by lightening (13) of their RFs. Redesigned from[42].…”

An Evolutionary Perspective on the Design of Neuromorphic Shape Filters

“…Therefore, the filter must provide a summary that is robust, allowing identification of the object using a minimal set of boundary markers. [33] It is entirely within the capacity of genes to pre-wire inborn shape filters. If artificial neural networks can provide for invariant discrimination of shapes with less than 100,000 training trails, the requisite connectivity could be selected by evolution over hundreds of millions of years.…”

“…Those visual skills appear to be available to fish, using the neural machinery of the retina and optic tectum. [23]- [33] These skills have substantial benefit, serving to increase the chances that a newborn (or newly hatched) animal will survive. It seems unlikely that they would be totally abandoned as new cortical tools were evolving.…”

Visual shape discrimination in goldfish, modelled with the neural circuitry of optic tectum and torus longitudinalis

“…The tectal layers, stratum opticum (SO) and stratum fibrosum et griseum superficiale (SFGS), receive most of the retinal ganglion cell terminals whose activity can be recorded with extracellular electrodes and distinguished from postsynaptic cell activity by various criteria. The most comprehensive investigations using these methods in goldfish tectum by the Maximovs and their associates revealed distinct functional types found at characteristic depths in SFGS (Maximova et al 2019). The units encountered most superficially were directionally selective, responding to edges moved in one of three directions.…”