Animals exhibit remarkable learning capabilities with little direct supervision. Likewise, self-supervised learning is an emergent paradigm in artificial intelligence, closing the performance gap to supervised learning. In the context of biology, self-supervised learning corresponds to a setting where one sense or specific stimulus may serve as a supervisory signal for another. After learning, the latter can be used to predict the former. On the implementation level, it has been demonstrated that such predictive learning can occur at the single neuron level, in compartmentalized neurons that separate and associate information from different streams. We demonstrate the power such self-supervised learning over unsupervised (Hebb-like) learning rules, which depend heavily on stimulus statistics, in two examples: First, in the context of animal navigation where predictive learning can associate internal self-motion information always available to the animal with external visual landmark information, leading to accurate path-integration in the dark. We focus on the well-characterized fly head direction system and show that our setting learns a connectivity strikingly similar to the one reported in experiments. The mature network is a quasi-continuous attractor and reproduces key experiments in which optogenetic stimulation controls the internal representation of heading, and where the network remaps to integrate with different gains. Second, we show that incorporating global gating by reward prediction errors allows the same setting to learn conditioning at the neuronal level with mixed selectivity. At its core, conditioning entails associating a neural activity pattern induced by an unconditioned stimulus (US) with the pattern arising in response to a conditioned stimulus (CS). Solving the generic problem of pattern-to-pattern associations naturally leads to emergent cognitive phenomena like blocking, overshadowing, saliency effects, extinction, interstimulus interval effects etc. Surprisingly, we find that the same network offers a reductionist mechanism for causal inference by resolving the post hoc, ergo propter hoc fallacy.

Neural responses are variable: even under identical experimental conditions, single neuron and population responses typically differ from trial to trial and across time. Recent work has demonstrated that this variability has predictable structure, can be modulated by sensory input and behaviour, and bears critical signatures of the underlying network dynamics and computations. However, current methods for characterising neural variability are primarily geared towards sensory coding in the laboratory: they require trials with repeatable experimental stimuli and behavioural covariates. In addition, they make strong assumptions about the parametric form of variability, rely on assumption-free but data-inefficient histogram-based approaches, or are altogether ill-suited for capturing variability modulation by covariates. Here we present a universal probabilistic spike count model that eliminates these shortcomings. Our method uses scalable Bayesian machine learning techniques to model arbitrary spike count distributions (SCDs) with flexible dependence on observed as well as latent covariates. Without requiring repeatable trials, it can flexibly capture covariate-dependent joint SCDs, and provide interpretable latent causes underlying the statistical dependencies between neurons. We apply the model to recordings from a canonical non-sensory neural population: head direction cells in the mouse. We find that variability in these cells defies a simple parametric relationship with mean spike count as assumed in standard models, its modulation by external covariates can be comparably strong to that of the mean firing rate, and slow low-dimensional latent factors explain away neural correlations. Our approach paves the way to understanding the mechanisms and computations underlying neural variability under naturalistic conditions, beyond the realm of sensory coding with repeatable stimuli.

The head direction (HD) system is classically modeled as a ring attractor network which ensures a stable representation of the animal’s head direction. This unidimensional description popularized the view of the HD system as the brain’s internal compass. However, unlike a globally consistent magnetic compass, the orientation of the HD system is dynamic, depends on local cues and exhibits remapping across familiar environments5. Such a system requires mechanisms to remember and align to familiar landmarks, which may not be well described within the classic 1-dimensional framework. To search for these mechanisms, we performed large population recordings of mouse thalamic HD cells using calcium imaging, during controlled manipulations of a visual landmark in a familiar environment. First, we find that realignment of the system was associated with a continuous rotation of the HD network representation. The speed and angular distance of this rotation was predicted by a 2nd dimension to the ring attractor which we refer to as network gain, i.e. the instantaneous population firing rate. Moreover, the 360-degree azimuthal profile of network gain, during darkness, maintained a ‘memory trace’ of a previously displayed visual landmark. In a 2nd experiment, brief presentations of a rotated landmark revealed an attraction of the network back to its initial orientation, suggesting a time-dependent mechanism underlying the formation of these network gain memory traces. Finally, in a 3rd experiment, continuous rotation of a visual landmark induced a similar rotation of the HD representation which persisted following removal of the landmark, demonstrating that HD network orientation is subject to experience-dependent recalibration. Together, these results provide new mechanistic insights into how the neural compass flexibly adapts to environmental cues to maintain a reliable representation of the head direction.

Tracking heading within an environment is a fundamental requirement for flexible, goal-directed navigation. In insects, a head-direction representation that guides the animal’s movements is maintained in a conserved brain region called the central complex. Two-photon calcium imaging of genetically targeted neural populations in the central complex of tethered fruit flies behaving in virtual reality (VR) environments has shown that the head-direction representation is updated based on self-motion cues and external sensory information, such as visual features and wind direction. Thus far, the head direction representation has mainly been studied in VR settings that only give flies control of the angular rotation of simple sensory cues. How the fly’s head direction circuitry enables the animal to navigate in dynamic, immersive and naturalistic environments is largely unexplored. I have developed a novel setup that permits imaging in complex VR environments that also accommodate flies’ translational movements. I have previously demonstrated that flies perform visually-guided navigation in such an immersive VR setting, and also that they learn to associate aversive optogenetically-generated heat stimuli with specific visual landmarks. A stable head direction representation is likely necessary to support such behaviors, but the underlying neural mechanisms are unclear. Based on a connectomic analysis of the central complex, I identified likely circuit mechanisms for prioritizing and combining different sensory cues to generate a stable head direction representation in complex, multimodal environments. I am now testing these predictions using calcium imaging in genetically targeted cell types in flies performing 2D navigation in immersive VR.

Many flexible behaviors are thought to rely on internal representations of an animal’s spatial relationship to its environment and of the consequences of its actions in that environment. While such representations—e.g. of head direction and value—have been extensively studied, how they are combined to guide behavior is not well understood. I will discuss how we are exploring these questions using a classical visual learning paradigm for the fly. I’ll begin by describing a simple policy that, when tethered to an internal representation of heading, captures structured behavioral variability in this task. I’ll describe how ambiguities in the fly’s visual surroundings affect its perception and, when coupled to this policy, manifest in predictable changes in behavior. Informed by newly-released connectomic data, I’ll then discuss how these computations might be carried out and combined within specific circuits in the fly’s central brain, and how perception and action might interact to shape individual differences in learning performance.