PUB - Publikationen an der Universität Bielefeld

How are external sensory stimuli perceived, integrated and represented within the central nervous system? How does the nervous system generate appropriate behavioral responses based on this input and how does this behavior affect perception?

The above questions have in common that they view sensory input and motor control as two sides of the sensorimotor loop. In this closed-loop system, actions inevitably generate sensory flow that can serve to organize behavior. To look, to smell, to touch, etc. are perceptual acts that depend on the interaction, coordination and interpretation of motor and sensory information through neural mechanisms. Active sensory systems are particularly amenable to the study of the reciprocal relations of motor and sensory components, as parts of closed loop control structures. A notable advantage of these sensory systems is the experimental accessibility of their sensory input, both in terms of its measurement and in terms of detailed modeling reconstructions of the input. In the case of weakly electric fish studied in this thesis, the animals sense and process environmental perturbations of a self-generated electric field. The fact that this field serves as the carrier of sensory information and at the same time is controlled by the animal, enables to precisely determine aspects of sensing that are often hard to obtain or quantify in sensory systems that do not actively generate the carrier: where, when and what an animal samples.
Drawing on these benefits, my thesis focuses on the role of motor and electromotor behavior in sensorimotor integration. For this, a biophysical model for the active and passive electroreception was combined with physiological recordings and behavioral approaches. The central topics addressed are:

(i) Object detection and sensorimotor learning. The sensory information obtained by the African species Gnathonemus petersii while learning a detection task was computationally reconstructed using boundary element methods (BEM). This revealed that the improved task performance was paralleled by an enhancement of the quality of the sensory information, which was mediated by changes of the electromotor patterns. The versatile manner in which the fish changed the spatial and temporal allocation of otherwise stable motor components not only improved the quality of the sensory input, but also resulted in shifts of the animals' attention towards the object.

(ii) Dynamic choice of optimal behavior. Extending on the above results, I next explored how changing the distance of an object to be detected by the fish influenced the electromotor behavior. With increasing complexity (distance), the fish resorted to a new motor strategy. This consisted in first approaching a salient element in the arena, from where the fish then made a perceptually-guided decision. This interpretation is backed up by analyzing the trajectories in the context of attractors, revealing that the focus of attention was altered in a task-dependent manner.

(iii) Distance estimation using a non-visual form of motion parallax. In the above experiments it is implicitly assumed that electric fish acquire spatial information like the position and distance of a target. How this is achieved dynamically has been addressed recently. Based on the properties of the electric field geometry, theoretical considerations indicated that relative movements might provide depth information. In a behavioral assay, I show that this novel form of electric parallax exists and is used across phylogenetically distant taxa of weakly electric fish (Apteronotus albifrons, Eigenmania virescens and Gnathonemus petersii). Notably, these species electrically sample the environment in temporally distinct ways (using discrete pulses or quasi-sinusoidal waves), suggesting an ubiquitous role for parallax in electric sensing.

(iv) The role of multi-modal integration in socially relevant agonistic behaviour. Extending on the above results, I next addressed if passive as well as active electric sensory information can be used to evaluate more complex features of the environment. For this I turned to social interactions of the South American species Gymnotus omarorum to study if an electrical assessment of a competitor is possible. Based on modeling the sensory consequences of dyadic encounters, I showed that passive as well as active sensory information can drive agonistic interactions. This suggests that aggressive interactions may be triggered by information about contenders obtained through the active and passive electrosensory system.

(v) Hierarchy as a social consequence of electric interactions. The above analysis indicated that active as well as passive electrolocation may contribute in a non-reciprocal manner to social interactions. Gymnotus omarorum then was tested in intra- and intersexual dyads in small plain arenas. A sex-independent dominant-subordinate status emerged after highly aggressive contests. Subordinates signaled their submission by retreating and emitting specific (submissive) electric signals. The emergence of a dominant-subordinate status was also observed in a larger arena after longer but milder contests with rare electric signaling of submission with a unique consequence: the persistence of dominance over time with no outcome reversion.