Population coding

Animals readily make value-based decisions between discrete and static items, with decades of work elucidating neural circuits involved in such choices (Rangel et al., 2015). Much less is known about naturalistic decision-making contexts such as goal pursuit (e.g., hunting). Dynamic choices require determining the value of blending different goal-pursuit strategies, switching goals, and continuously monitoring the actions and goals most valuable (Yoo et al., 2021). Despite its prevalence, there is a lack of methods to identify continuous decision processes and thus a dearth of understanding th

Predicting neural activity in higher visual areas (HVAs) in mice during spatial navigation is challenging due to the complex interaction between spatial and temporal processing. One recent two-stream hypothesis for rodents proposed by Saleem (2020) suggests that form-based goals and detailed landmark signals are processed by cortical areas biased towards the central visual field, while self-motion signals are processed in the periphery. Inspired by this, we developed a deep learning model that splits visual input into two streams: a convolutional neural network (CNN) processed the central visu

Spatial navigation relies on external cues, such as visual objects, to anchor internal representations of space. In rodents, moving a visual landmark results in a corresponding shift in the tuning of head direction (HD) cells1. However, little is known about where visual objects are encoded in the mouse brain2 or how visual objects modulate the firing properties of HD cells3. Using functional ultrasound (fUS) imaging4, we conducted a brain-wide screen to identify areas preferentially activated by visual objects compared to scrambled versions of the same stimuli. Our results showed that while v

…such as hierarchical predictive coding[1] or hierarchical Bayesian inference[2] postulate abstract normative goals for the role of feedback connections. However, predictions derived from these theories are typically specified at the level of individual neurons. Whether feedback shapes collective properties of neural populations and how such changes may impact encoding of sensory signals remain unknown. Here we tackle these questions directly by studying the population coding of na

The population coding account of pain proposes that a distributed network of neural activity across both nociceptive and non-nociceptive domains drives our experience of thermosensation and pain (Coghill, 2020; Fardo et al., 2020). Recent, pioneering work in mice shows that distinct neuronal population codes for cold and warm exist in the posterior insula cortex (Vestergaard et al., 2023). In this study we used fMRI to model the pattern of response associated with innocuous and noxious thermal stimuli, to determine whether similar population receptive fields (PRFs) for temperature and pain exi

Learning relies as much on synaptic weight modifications as it does on the timing of neural signaling. For instance, variability in axonal length and diameter may compromise spike-timing-dependent plasticity (STDP) by dispersing action potential timing, preventing synaptic potentiation and depression. Activity-dependent myelination (ADM), a form of brain plasticity relying on neuron-glia feedback, has been shown to normalize conduction times across different axons through adaptive, neuron-activity-dependent changes in axonal conduction velocity (CV). While historically seen as a mere compensat

…to elegant, interpretable models of population coding and working memory. Frequently, attractors are classified based on whether their invariant manifolds can represent discrete/continuous variables, with point attractors being an example of the former and ‘continuous attractors’ of the latter. Here we highlight that ‘learnability’ is another quality that may serve to differentiate neural dynamics while offering novel insight. Counterintuitively, good working memory does not necessarily imply that a well-de

Causal inference based on temporal relationships is fundamental for adaptive, intelligent behavior. However, standard value-based reinforcement learning (RL) models learn estimates of temporally discounted average future reward, leading to ambiguity about future reward timing and magnitude. Extending these algorithms to learn a set of values that differ in their sensitivity to reward prediction errors (RPEs) and temporal discounting can support inference. Midbrain dopamine (DA) neurons, known to encode RPEs, display signatures of learning about distributions of reward magnitude. However, it is

Norepinephrine (NE) is a neuromodulator that is released from projections of the locus coeruleus via extra-synaptic vesicle exocytosis (Fuxe, Borroto-Escuela, front. Physiol 2012). Previously, NE in the prefrontal cortex (PFC) was thought to be a homogenous field created by bulk release, but it remains unknown whether phasic (fast, short-term) fluctuations in NE can produce a spatially heterogeneous field, which could then structure cell firing at a fine spatial scale. To understand how spatiotemporal dynamics of NE release in the PFC affect neuronal firing, we performed a novel in-vivo two-ph

…and test a theory of hippocampal coding which produces place cells as a general consequence of

To localize ourselves in an environment for spatial navigation, we rely on vision and self-motion inputs, which only provide noisy and partial information. It is unknown how the resulting uncertainty affects navigation behavior and neural representations. Here we show that spatial uncertainty underlies key effects of environmental geometry on navigation behavior and grid field deformations. We develop an ideal observer model, which continually updates probabilistic beliefs about its allocentric location by optimally combining noisy egocentric visual and self-motion inputs via Bayesian filterin

The cerebellum resembles a feedforward, three-layer network of neurons in which the “hidden layer” consists of Purkinje cells (P-cells) and the output layer consists of deep cerebellar nucleus (DCN) neurons. In this analogy, the output of each DCN neuron is a prediction that is compared with the actual observation, resulting in an error signal that originates in the inferior olive. Efficient learning requires that the error signal reach the DCN neurons, as well as the P-cells that project onto them. However, this basic rule of learning is violated in the cerebellum: the olivary projections to

Representational geometry is a prevalent approach to abstract and investigate population coding derived from the activities of individual neurons. However, the same representational geometry can be implemented by neurons in different ways, which implies different potential readout mechanisms. Leveraging the biological constraint of sparsity of readout synapses, we identified the biologically meaningful sparse components that express the geometry of working memory representations in primate prefrontal cortex (PFC). The dominant neuron groups corresponding to each component had distinct electrop

…and from overlapping neural populations. Here, we passively exposed anesthetized rats to complex auditory sequences consisting of artificial vowel triple