Perceptual decisions, learned and innate
Animals must make a multitude of decisions based on incoming sensory input, ranging from the navigational (which path will lead to a target location) to the categorical (which object most reliably leads to reward). I study how animals make these perceptual decisions, developing custom behavioral assays to probe the strategies used, and using electrophysiology and optical calcium imaging to understand how the brain implements these strategies.
In my previous work, I used a visual decision task to ask how rats process information across their visual field. We found that V1 neurons carried information about the visual, choice, and reward features of the task in a distributed manner. We were also able to use targeted probe stimuli in this task to both discover and modulate the decision strategy of the rats.
My current project asks how fish make swim decisions on the basis of visual and mechanosensory stimuli that provide information about the fluid flow of their environment. Here, we leverage the innate drive of the larval zebrafish to swim against the flow, and use both naturalistic and constrained custom behavioral setups to elicit and measure swim responses. We combine analysis of fish behavior at the level of single swim decisions with two-photon calcium imaging to understand the underlying circuit computations.
Flow navigation as a model of multisensory processing
Because larval zebrafish hatch only a couple of days after fertilization, they possess a collection of robust innate behaviors that are crucial for survival. These include escape responses, hunting sequences, and importantly, “reflexes” that help them to stabilize their position in the face of flow.
These reflexes are described for a unimodal input, but flow information is typically multimodal, with visual (retinal slip as the fish is dragged) and mechanosensory (bending of the lateral line hair cells as fluid moves over the fish) components. As mentioned above, we have developed behavioral setups that can robustly deliver both of these stimuli simultaneously to the fish, and we use them to ask how fish use these two streams of information to guide their response to flow fields. We ask what happens when these two information streams go into conflict, and whether there is any adaptation or time-varying component of fish navigation decisions.
Fin-tail coordination in larval zebrafish
Inferring the internal state or behavioral strategy of an animal is difficult, and is made more challenging when measurable outputs from the animal are low dimensional. In the larval zebrafish, the routine swims and turns that fish use to respond to environmental stimuli such as flow are summarized by their resultant angular propulsion and vigor. Concurrent instantaneous readouts can provide more insight into momentary internal state; for example, pupil diameter in rodents can serve as a measure of arousal.
In fish, we ask whether we can measure fine adjustments of motor effectors including and beyond the tail, and whether these measurements can provide additional information about the state of the fish. In particular, the pectoral fins are typically thought to be ineffective for propulsion at this age, but are always engaged during swim bouts. We are investigating how the movement of pectoral fins is coordinated with tail movement across different stimulus conditions, and whether these two sets of effectors carry fully overlapping or partially independent information about the motor state (and potentially internal state) of the fish. We have developed a multicamera setup that allows us to track the movement of these effectors in three dimensional space, and gain access to elevational or rotational axes of motor control that would be missed from the classical top-down view.
High-throughput behavioral screening across genotypes and treatments
Because flow navigation is both a core component of the larval zebrafish’s behavioral repertoire, and a flexible decision process where the fish can modulate vigor, orientation, engagement, and sensory weights, it serves as both a robust and sensitive measure of an animal’s behavioral state. We use our naturalistic free swimming arena to track the swim statistics of different species and genotypes of fish, and to measure the effects of drug treatments. We can measure the positions of multiple fish simultaneously, thereby allowing us to use flow navigation in this arena as a screening tool for groups of fish.