Select Publications

An Internal Model for Canceling Self-Generated Sensory Input in Freely Behaving Electric Fish

The brain predicts and corrects for how your own movement affects the visual input you receive from moment to moment. Electric fish navigate, communicate and track prey with organs specialized for generating and detecting electric fields. To survive, they need to distinguish changes in the fields they emit due to their own bodily movements from those associated with nearby plants, fish and the water fleas they eat. We studied fish swimming freely in a round tank and monitored brain activity within the creatures’ electrosensory lobe (ELL), a region that integrates motor signals with various sensory signals, to filter-out sensory noise. The data shows a sophisticated filtering process which adapts to the animals' actions as well as to its surrounding environment.

Wallach & Sawtell, Neuron (2023)

The brain translates complex scenarios into neural code. Many recent studies showed ‘mixed selectivity coding’, in which each neuron simultaneously conveys information on multiple aspects of the world. This was mostly tested in limited artificial lab scenarios. We simulated social interactions between electric fish, while recording neurons in the fish’s Thalamus – a ‘bottleneck’ connecting its low-level and high-level brain circuits. When two fish are close to one another, their electric fields interact to create an interference pattern. The pattern’s frequency informs them of each other’s identity (Who?) and its amplitude- of their relative position and movement (Where?). On the background of this ‘channel’ the fish exchange messages, called ‘chirps’, to signal courtship, aggression, submission, etc. Thalamic cells responded to mixtures of these features: each cell was active in the presence of fish of a particular identity (e.g., same-sex or opposite sex), with a particular relative motion (e.g. approaching or retreating), and when particular chirps occurred. With this understanding of how natural social scenes are translated into neural code, we can develop hypotheses about how the high-level circuits in the forebrain interpret and control such behavioral interactions

Wallach, Melanson, Longtin, & Maler, Current Biology (2022)

A time-stamp mechanism may provide temporal information necessary for egocentric to allocentric spatial transformations

Finding their way around is an essential part of survival for many animals and helps them to locate food, mates and shelter. Animals have evolved the ability to form a 'map' or representation of their surroundings. For example, the electric fish Apteronotus leptorhynchus, can precisely learn the location of food and navigate there in complete darkness using the weak electric field it generates. As it swims, every object it encounters generates an ‘electric image’ that is detected on the skin and processed in the brain. However, all the cues the fish comes across are from its own point of view – the information about its environment is processed with respect to its location. The way animals translate ‘self-centered’ experiences to form a general representation of their surroundings is not yet fully understood. Information about the fish's environment passes through a structure in the brain called the preglomerular complex. Measuring the activity of this region revealed that instead of processing self-centered information, preglomerular activity is generated every time the fish encounters external objects. The intensity of the activity depended on how recently the fish had encountered another object. This information, combined with the dynamics of the fish's movement, could be what allows the fish to convert a sequence of encounters into a general spatial map.

Wallach, Harvey-Girard, Jun, J. J., Longtin, & Maler, Elife (2018)

Predictive whisker kinematics reveal context-dependent sensorimotor strategies 

Animals actively move their sensory organs in order to acquire sensory information. We show that active sensing behavior is affected by the behavioral context determined by both internal processes (attention and expectations) and external constraints (available sensory and motor degrees of freedom). This work shows that rats adapt their active exploratory behavior in a homeostatic attempt to preserve sensorimotor coverage under changing environmental conditions and changing sensory capacities, including those imposed by various laboratory conditions.

Wallach, Deutsch, Oram, & Ahissar, PLOS Biology (2020)

On-going computation of whisking phase by mechanoreceptors

Using a closed-loop rat-machine “hybrid” system that uses real-time motor control to generate awake-like whisker motion in anesthetized rats, we systematically characterized the response of the whisker mechanoreceptors and their brainstem targets to naturalistic behavior. We discovered that the receptors perform a predictive computation that extracts the whisking phase from the whisker kinematics. This work was awarded me the Society for Neuroethology’s Young Investigator Award (received at the Society’s biannual meeting of 2016 in Montevideo, Uruguay).

Wallach, Bagdassarian & Ahissar, Nature Neuroscience (2016)