Tuning movement for sensing in an uncertain world

Sensorimotor adaptation to destabilizing dynamics in weakly electric fish

Humans and other animals can readily learn to compensate for changes in the dynamics of movement. Such changes can result from an injury or changes in the weight of carried objects. These changes in dynamics can lead not only to reduced performance but also to dramatic instabilities. We evaluated the impacts of compensatory changes in control policies in relation to stability and robustness in Eigenmannia virescens, a species of weakly electric fish. We discovered that these fish retune their sensorimotor control system in response to experimentally generated destabilizing dynamics. Specifically, we used an augmented reality system to manipulate sensory feedback during an image stabilization task in which a fish maintained its position within a refuge. The augmented reality system measured the fish's movements in real time. These movements were passed through a high-pass filter and multiplied by a gain factor before being fed back to the refuge motion. We adjusted the gain factor to gradually destabilize the fish's sensorimotor loop. The fish retuned their sensorimotor control system to compensate for the experimentally induced destabilizing dynamics. This retuning was partially maintained when the augmented reality feedback was abruptly removed. The compensatory changes in sensorimotor control improved tracking performance as well as control-theoretic measures of robustness, including reduced sensitivity to disturbances and improved phase margins.

Behavioral Motor Performance

The human sensorimotor control system has exceptional abilities to perform skillful actions. We easily switch between strenuous tasks that involve brute force, such as lifting a heavy sewing machine, and delicate movements such as threading a needle in the same machine. Using a structure with different control architectures, the motor system is capable of updating its ability to perform through our daily interaction with the fluctuating environment. However, there are issues that make this a difficult computational problem for the brain to solve. The brain needs to control a nonlinear, nonstationary neuromuscular system, with redundant and occasionally undesired degrees of freedom, in an uncertain environment using a body in which information transmission is subject to delays and noise. To gain insight into the mechanisms of motor control, here we survey movement laws and invariances that shape our everyday motion. We then examine the major solutions to each of these problems in the three parts of the sensorimotor control system, sensing, planning, and acting. We focus on how the sensory system, the control architectures, and the structure and operation of the muscles serve as complementary mechanisms to overcome deviations and disturbances to motor behavior and give rise to skillful motor performance. We conclude with possible future research directions based on suggested links between the operation of the sensorimotor system across the movement stages. © 2024 American Physiological Society. Compr Physiol 14:5179-5224, 2024.

Active Sampling in Primate Vocal Interactions

Active sensing is a behavioral strategy for exploring the environment. In this study, we show that contact vocal behaviors can be an active sensing mechanism that uses sampling to gain information about the social environment, in particular, the vocal behavior of others. With a focus on the realtime vocal interactions of marmoset monkeys, we contrast active sampling to a vocal accommodation framework in which vocalizations are adjusted simply to maximize responses. We conducted simulations of a vocal accommodation and an active sampling policy and compared them with real vocal exchange data. Our findings support active sampling as the best model for marmoset monkey vocal exchanges. In some cases, the active sampling model was even able to predict the distribution of vocal durations for individuals. These results suggest a new function for primate vocal interactions in which they are used by animals to seek information from social environments.

Active smelling in the American cockroach

Motion plays an essential role in sensory acquisition. From changing the position in which information can be acquired to fine-scale probing and active sensing, animals actively control the way they interact with the environment. In olfaction, movement impacts the time and location of odour sampling as well as the flow of odour molecules around the olfactory organs. Employing a detailed spatiotemporal analysis, we investigated how insect antennae interact with the olfactory environment in a species with a well-studied olfactory system - the American cockroach. Cockroaches were tested in a wind-tunnel setup during the presentation of odours with different attractivity levels: colony extract, butanol and linalool. Our analysis revealed significant changes in antennal kinematics when odours were presented, including a shift towards the stream position, an increase in vertical movement and high-frequency local oscillations. Nevertheless, the antennal shifting occurred predominantly in a single antenna while the overall range covered by both antennae was maintained throughout. These findings hold true for both static and moving stimuli and were more pronounced for attractive odours. Furthermore, we found that upon odour encounter, there was an increase in the occurrence of high-frequency antennal sweeps and vertical strokes, which were shown to impact the olfactory environment's statistics directly. Our study lays out a tractable system for exploring the tight coupling between sensing and movement, in which antennal sweeps, in parallel to mammalian sniffing, are actively involved in facilitating odour capture and transport, generating odour intermittency in environments with low air movement where cockroaches dwell.

Active Sensing in Bees Through Antennal Movements Is Independent of Odor Molecule

When sampling odors, many insects are moving their antennae in a complex but repeatable fashion. Previous studies with bees have tracked antennal movements in only two dimensions, with a low sampling rate and with relatively few odorants. A detailed characterization of the multimodal antennal movement patterns as function of olfactory stimuli is thus wanted. The aim of this study is to test for a relationship between the scanning movements and the properties of the odor molecule. We tracked several key locations on the antennae of bumblebees at high frequency and in three dimensions while stimulating the insect with puffs of 11 common odorants released in a low-speed continuous flow. Water and paraffin were used as negative controls. Movement analysis was done with the neural network Deeplabcut. Bees use a stereotypical oscillating motion of their antennae when smelling odors, similar across all bees, independently of the identity of the odors and hence their diffusivity and vapor pressure. The variability in the movement amplitude among odors is as large as between individuals. The main type of oscillation at low frequencies and large amplitude is triggered by the presence of an odor and is in line with previous work, as is the speed of movement. The second oscillation mode at higher frequencies and smaller amplitudes is constantly present. Antennae are quickly deployed when a stimulus is perceived, decorrelate their movement trajectories rapidly, and oscillate vertically with a large amplitude and laterally with a smaller one. The cone of airspace thus sampled was identified through the 3D understanding of the motion patterns. The amplitude and speed of antennal scanning movements seem to be function of the internal state of the animal, rather than determined by the odorant. Still, bees display an active olfactory sampling strategy. First, they deploy their antennae when perceiving an odor. Second, fast vertical scanning movements further increase the odorant capture rate. Finally, lateral movements might enhance the likelihood to locate the source of odor, similarly to the lateral scanning movement of insects at odor plume boundaries.

Finding food in the dark: how trajectories of a gymnotiform fish change with spatial learning

We analyzed the trajectories of freely foraging Gymnotus sp., a pulse-type gymnotiform weakly electric fish, swimming in a dark arena. For each fish, we compared the its initial behavior as it learned the relative location of landmarks and food with its behavior after learning was complete, i.e. after time/distance to locate food had reached a minimal asymptotic level. During initial exploration when the fish did not know the arena layout, trajectories included many sharp angle head turns that occurred at nearly completely random intervals. After spatial learning was complete, head turns became far smoother. Interestingly, the fish still did not take a stereotyped direct route to the food but instead took smooth but variable curved trajectories. We also measured the fish's heading angle error (heading angle - heading angle towards food). After spatial learning, the fish's initial heading angle errors were strongly biased to zero, i.e. the fish mostly turned towards the food. As the fish approached closer to the food, they switched to a random search strategy with a more uniform distribution of heading angle errors.

Mechanical intelligence for learning embodied sensor-object relationships

Intelligence involves processing sensory experiences into representations useful for prediction. Understanding sensory experiences and building these contextual representations without prior knowledge of sensor models and environment is a challenging unsupervised learning problem. Current machine learning methods process new sensory data using prior knowledge defined by either domain knowledge or datasets. When datasets are not available, data acquisition is needed, though automating exploration in support of learning is still an unsolved problem. Here we develop a method that enables agents to efficiently collect data for learning a predictive sensor model-without requiring domain knowledge, human input, or previously existing data-using ergodicity to specify the data acquisition process. This approach is based entirely on data-driven sensor characteristics rather than predefined knowledge of the sensor model and its physical characteristics. We learn higher quality models with lower energy expenditure during exploration for data acquisition compared to competing approaches, including both random sampling and information maximization. In addition to applications in autonomy, our approach provides a potential model of how animals use their motor control to develop high quality models of their sensors (sight, sound, touch) before having knowledge of their sensor capabilities or their surrounding environment.

Searching for a source without gradients: how good is infotaxis and how to beat it

Infotaxis is a popular search algorithm designed to track a source of odour in a turbulent environment using information provided by odour detections. To exemplify its capabilities, the source-tracking task was framed as a partially observable Markov decision process consisting in finding, as fast as possible, a stationary target hidden in a two-dimensional grid using stochastic partial observations of the target location. Here, we provide an extended review of infotaxis, together with a toolkit for devising better strategies. We first characterize the performance of infotaxis in domains from one dimension to four dimensions. Our results show that, while being suboptimal, infotaxis is reliable (the probability of not reaching the source approaches zero), efficient (the mean search time scales as expected for the optimal strategy) and safe (the tail of the distribution of search times decays faster than any power law, though subexponentially). We then present three possible ways of beating infotaxis, all inspired by methods used in artificial intelligence: tree search, heuristic approximation of the value function, and deep reinforcement learning. The latter is able to find, without any prior human knowledge, the (near) optimal strategy. Altogether, our results provide evidence that the margin of improvement of infotaxis towards the optimal strategy gets smaller as the dimensionality increases.

A mechanism for punctuating equilibria during mammalian vocal development

Evolution and development are typically characterized as the outcomes of gradual changes, but sometimes (states of equilibrium can be punctuated by sudden change. Here, we studied the early vocal development of three different mammals: common marmoset monkeys, Egyptian fruit bats, and humans. Consistent with the notion of punctuated equilibria, we found that all three species undergo at least one sudden transition in the acoustics of their developing vocalizations. To understand the mechanism, we modeled different developmental landscapes. We found that the transition was best described as a shift in the balance of two vocalization landscapes. We show that the natural dynamics of these two landscapes are consistent with the dynamics of energy expenditure and information transmission. By using them as constraints for each species, we predicted the differences in transition timing from immature to mature vocalizations. Using marmoset monkeys, we were able to manipulate both infant energy expenditure (vocalizing in an environment with lighter air) and information transmission (closed-loop contingent parental vocal playback). These experiments support the importance of energy and information in leading to punctuated equilibrium states of vocal development.

Species can sometimes evolve suddenly; their appearance is preceded and followed by long periods of stability. This process is known as “punctuated equilibrium”. Our data show that for three mammalian species—marmoset monkeys, fruit bats, and humans—early vocal development trajectories can also be characterized as different equilibrium states punctuated by sharp transitions; transitions indicate the advent of a new vocal behavior. To better understand the putative mechanism behind such transitions, we show that a balance model, in which variables trade-off in their importance over time, captured this change by accurately simulating the shape of the developmental trajectory and predicting the timing of the transition between immature and mature vocal states for all three species. Two variables—energy and information—were hypothesized to trade-off during development. We tested and found support for this hypothesis in analyses of two marmoset monkey experiments, one which manipulated energy metabolic costs and another which manipulated information transmission.

Formalizing planning and information search in naturalistic decision-making

Decisions made by mammals and birds are often temporally extended. They require planning and sampling of decision-relevant information. Our understanding of such decision-making remains in its infancy compared with simpler, forced-choice paradigms. However, recent advances in algorithms supporting planning and information search provide a lens through which we can explain neural and behavioral data in these tasks. We review these advances to obtain a clearer understanding for why planning and curiosity originated in certain species but not others; how activity in the medial temporal lobe, prefrontal and cingulate cortices may support these behaviors; and how planning and information search may complement each other as means to improve future action selection.