Publications


2021

2020

2019

2018

2013-2017

2011

2009 – 2010

 

.

2021

Abstract: Hippocampal neurons encode physical variables1,2,3,4,5,6,7 such as space1 or auditory frequency6 in cognitive maps8. In addition, functional magnetic resonance imaging studies in humans have shown that the hippocampus can also encode more abstract, learned variables9,10,11. However, their integration into existing neural representations of physical variables12,13 is unknown. Here, using two-photon calcium imaging, we show that individual neurons in the dorsal hippocampus jointly encode accumulated evidence with spatial position in mice performing a decision-making task in virtual reality14,15,16. Nonlinear dimensionality reduction13 showed that population activity was well-described by approximately four to six latent variables, which suggests that neural activity is constrained to a low-dimensional manifold. Within this low-dimensional space, both physical and abstract variables were jointly mapped in an orderly manner, creating a geometric representation that we show is similar across mice. The existence of conjoined cognitive maps suggests that the hippocampus performs a general computation—the creation of task-specific low-dimensional manifolds that contain a geometric representation of learned knowledge.

Context-based sensorimotor routing is a hallmark of executive control. Pharmacological inactivations in rats have implicated the midbrain superior colliculus (SC) in this process. But what specific role is this, and what circuit mechanisms support it? Here we report a subset of rat SC neurons that instantiate a specific link between the representations of context and motor choice. Moreover, these neurons encode animals’ choice far earlier than other neurons in the SC or in the frontal cortex, suggesting that their neural dynamics lead choice computation. Optogenetic inactivations revealed that SC activity during context encoding is necessary for choice behavior, even while that choice behavior is robust to inactivations during choice formation. Searches for SC circuit models matching our experimental results identified key circuit predictions while revealing some a priori expected features as unnecessary. Our results reveal circuit mechanisms within the SC that implement response inhibition and context-based vector inversion during executive control.

Decision-making strategies evolve during training and can continue to vary even in well-trained animals. However, studies of sensory decision-making tend to characterize behavior in terms of a fixed psychometric function that is fit only after training is complete. Here, we present PsyTrack, a flexible method for inferring the trajectory of sensory decision-making strategies from choice data. We apply PsyTrack to training data from mice, rats, and human subjects learning to perform auditory and visual decision-making tasks. We show that it successfully captures trial-to-trial fluctuations in the weighting of sensory stimuli, bias, and task-irrelevant covariates such as choice and stimulus history. This analysis reveals dramatic differences in learning across mice and rapid adaptation to changes in task statistics. PsyTrack scales easily to large datasets and offers a powerful tool for quantifying time-varying behavior in a wide variety of animals and tasks.

.

2020

How does the brain internally represent a sequence of sensory information that jointly drives a decision-making behavior? Studies of perceptual decision-making have often assumed that sensory cortices provide noisy but otherwise veridical sensory inputs to downstream processes that accumulate and drive decisions. However, sensory processing in even the earliest sensory cortices can be systematically modified by various external and internal contexts. We recorded from neuronal populations across posterior cortex as mice performed a navigational decision-making task based on accumulating randomly timed pulses of visual evidence. Even in V1, only a small fraction of active neurons had sensory-like responses time-locked to each pulse. Here, we focus on how these ‘cue-locked’ neurons exhibited a variety of amplitude modulations from sensory to cognitive, notably by choice and accumulated evidence. These task-related modulations affected a large fraction of cue-locked neurons across posterior cortex, suggesting that future models of behavior should account for such influences.

The use of Neuropixels probes for chronic neural recordings is in its infancy and initial studies leave questions about long-term stability and probe reusability unaddressed. Here, we demonstrate a new approach for chronic Neuropixels recordings over a period of months in freely moving rats. Our approach allows multiple probes per rat and multiple cycles of probe reuse. We found that hundreds of units could be recorded for multiple months, but that yields depended systematically on anatomical position. Explanted probes displayed a small increase in noise compared to unimplanted probes, but this was insufficient to impair future single-unit recordings. We conclude that cost-effective, multi-region, and multi-probe Neuropixels recordings can be carried out with high yields over multiple months in rats or other similarly sized animals. Our methods and observations may facilitate the standardization of chronic recording from Neuropixels probes in freely moving animals.

.

2019

We develop a phenomenological coarse-graining procedure for activity in a large network of neurons, and apply this to recordings from a population of 1000+ cells in the hippocampus. Distributions of coarse-grained variables seem to approach a fixed non-Gaussian form, and we see evidence of scaling in both static and dynamic quantities. These results suggest that the collective behavior of the network is described by a nontrivial fixed point.

Individual choices are not made in isolation but are embedded in a series of past experiences, decisions, and outcomes. The effects of past experiences on choices, often called sequential biases, are ubiquitous in perceptual and value-based decision-making, but their neural substrates are unclear. We trained rats to choose between cued guaranteed and probabilistic rewards in a task in which outcomes on each trial were independent. Behavioral variability often reflected sequential effects, including increased willingness to take risks following risky wins, and spatial ‘win-stay/lose-shift’ biases. Recordings from lateral orbitofrontal cortex (lOFC) revealed encoding of reward history and receipt, and optogenetic inhibition of lOFC eliminated rats’ increased preference for risk following risky wins, but spared other sequential effects. Our data show that different sequential biases are neurally dissociable, and the lOFC’s role in adaptive behavior promotes learning of more abstract biases (here, biases for the risky option), but not spatial ones.

In 1979, Daniel Kahneman and Amos Tversky published a ground-breaking paper titled “Prospect Theory: An Analysis of Decision under Risk,” which presented a behavioral economic theory that accounted for the ways in which humans deviate from economists’ normative workhorse model, Expected Utility Theory. For example, people exhibit probability distortion (they overweight low probabilities), loss aversion (losses loom larger than gains), and reference dependence (outcomes are evaluated as gains or losses relative to an internal reference point). We found that rats exhibited many of these same biases, using a task in which rats chose between guaranteed and probabilistic rewards. However, prospect theory assumes stable preferences in the absence of learning, an assumption at odds with alternative frameworks such as animal learning theory and reinforcement learning. Rats also exhibited trial history effects, consistent with ongoing learning. A reinforcement learning model in which state-action values were updated by the subjective value of outcomes according to prospect theory reproduced rats’ nonlinear utility and probability weighting functions and also captured trial-by-trial learning dynamics.

Neural activity throughout the cortex is correlated with perceptual decisions, but inactivation studies suggest that only a small number of areas are necessary for these behaviors. Here we show that the number of required cortical areas and their dynamics vary across related tasks with different cognitive computations. In a visually guided virtual T-maze task, bilateral inactivation of only a few dorsal cortical regions impaired performance. In contrast, in tasks requiring evidence accumulation and/or post-stimulus memory, performance was impaired by inactivation of widespread cortical areas with diverse patterns of behavioral deficits across areas and tasks. Wide-field imaging revealed widespread ramps of Ca2+activity during the accumulation and visually guided tasks. Additionally, during accumulation, different regions had more diverse activity profiles, leading to reduced inter-area correlations. Using a modular recurrent neural network model trained to perform analogous tasks, we argue that differences in computational strategies alone could explain these findings.

.

2018

Widefield imaging of calcium dynamics is an emerging method for mapping regional neural activity but is currently limited to restrained animals. Here we describe cScope, a head-mounted widefield macroscope developed to image large-scale cortical dynamics in rats during natural behavior. cScope provides a 7.8 × 4 mm field of view and dual illumination paths for both fluorescence and hemodynamic correction and can be fabricated at low cost using readily attainable components. We also report the development of Thy-1 transgenic rat strains with widespread neuronal expression of the calcium indicator GCaMP6f. We combined these two technologies to image large-scale calcium dynamics in the dorsal neocortex during a visual evidence accumulation task. Quantitative analysis of task-related dynamics revealed multiple regions having neural signals that encode behavioral choice and sensory evidence. Our results provide a new transgenic resource for calcium imaging in rats and extend the domain of head-mounted microscopes to larger-scale cortical dynamics.

Decision making in dynamic environments requires discounting old evidence that may no longer inform the current state of the world. Previous work found that humans discount old evidence in a dynamic environment, but do not discount at the optimal rate. Here we investigated whether rats can optimally discount evidence in a dynamic environment by adapting the timescale over which they accumulate evidence. Using discrete evidence pulses, we exactly compute the optimal inference process. We show that the optimal timescale for evidence discounting depends on both the stimulus statistics and noise in sensory processing. When both of these components are taken into account, rats accumulate and discount evidence with the optimal timescale. Finally, by changing the volatility of the environment, we demonstrate experimental control over the rats’ accumulation timescale. The mechanisms supporting integration are a subject of extensive study, and experimental control over these timescales may open new avenues of investigation.

A broad range of decision-making processes involve gradual accumulation of evidence over time, but the neural circuits responsible for this computation are not yet established. Recent data indicate that cortical regions that are prominently associated with accumulating evidence, such as the posterior parietal cortex and the frontal orienting fields, may not be directly involved in this computation. Which, then, are the regions involved? Regions that are directly involved in evidence accumulation should directly influence the accumulation-based decision-making behavior, have a graded neural encoding of accumulated evidence and contribute throughout the accumulation process. Here, we investigated the role of the anterior dorsal striatum (ADS) in a rodent auditory evidence accumulation task using a combination of behavioral, pharmacological, optogenetic, electrophysiological and computational approaches. We find that the ADS is the first brain region known to satisfy the three criteria. Thus, the ADS may be the first identified node in the network responsible for evidence accumulation.

We describe a mouse decision-making task based on accumulation of visual pulses of evidence, while the mice navigate a T-maze in virtual reality. This is done in head-fixed mice, allowing the cognitive task to thus serve as a platform for two-photon microscopy, laser-scanning inactivations, and other methods requiring head fixation.

Many models of cognition and of neural computations posit the use and estimation of prior stimulus statistics: it has long been known that working memory and perception are strongly impacted by previous sensory experience, even when that sensory history is not relevant to the current task at hand. Nevertheless, the neural mechanisms and regions of the brain that are necessary for computing and using such prior experience are unknown. Here we report that the posterior parietal cortex (PPC) is a critical locus for the representation and use of prior stimulus information. We trained rats in an auditory parametric working memory task, and found that they displayed substantial and readily quantifiable behavioural effects of sensory-stimulus history, similar to those observed in humans and monkeys. Earlier proposals that the PPC supports working memory predict that optogenetic silencing of this region would impair behaviour in our working memory task. Contrary to this prediction, we found that silencing the PPC significantly improved performance. Quantitative analyses of behaviour revealed that this improvement was due to the selective reduction of the effects of prior sensory stimuli. Electrophysiological recordings showed that PPC neurons carried far more information about the sensory stimuli of previous trials than about the stimuli of the current trial. Furthermore, for a given rat, the more information about previous trial sensory history in the neural firing rates of the PPC, the greater the behavioural effect of sensory history, suggesting a tight link between behaviour and PPC representations of stimulus history. Our results indicate that the PPC is a central component in the processing of sensory-stimulus history, and could enable further neurobiological investigation of long-standing questions regarding how perception and working memory are affected by prior sensory information.


.

2013-2017

Planning can be defined as action selection that leverages an internal model of the outcomes likely to follow each possible action. Its neural mechanisms remain poorly understood. Here we adapt recent advances from human research for rats, presenting for the first time an animal task that produces many trials of planned behavior per session, making multitrial rodent experimental tools available to study planning. We use part of this toolkit to address a perennially controversial issue in planning: the role of the dorsal hippocampus. Although prospective hippocampal representations have been proposed to support planning, intact planning in animals with damaged hippocampi has been repeatedly observed. Combining formal algorithmic behavioral analysis with muscimol inactivation, we provide causal evidence directly linking dorsal hippocampus with planning behavior. Our results and methods open the door to new and more detailed investigations of the neural mechanisms of planning in the hippocampus and throughout the brain.

Previous work from the lab has shown that perturbing the FOF during decision-making tasks produces a difficulty-independent bias. We sought to investigate what computational role for the FOF could produce such a bias. We found that when a simple two-node attractor network was asked to accumulate evidence or categorize already accumulated evidence into decisions, they could not fit the experimental data well. However, when the network was asked to remember an already formed decision, the networks could match the data very well. This network in a “post-categorization” role naturally accounts for a variety of other data from the lab about FOF tuning curves, as well as behavioral responses to fast timescale inactivations, reinforcing the idea that the FOF is more involved in maintaining the memory of a decision than in forming the decision itself.

Ben and Christine recorded from frontal and parietal cortex using two-photon calcium imaging techniques while rats performed a pulse-based accumulation of evidence task.  They found that each pulse of evidence triggered a wave of activity that slowly spread through the cortex. These waves continued to propagate throughout the cortex until the choice was made. Although the response of each individual neuron was relatively brief, together they contained enough information to accurately predict the number of light pulses that the animal saw. The activity may reflect a strategy to construct memories that are stable over time but that can also be easily modified as new information becomes available.

Many models of decision-making process include an “evidence accumulation” stage: evidence for or against different options is gradually accumulated over time, and it is the result of that accumulation process that drives the final decision. In these models (including work from our own lab), it is usually assumed that noise from inputs at different timepoints is independent of each other, in which case the total variance would increase linearly with the number of inputs. But is that true? Using a visual flashes accumulation task, we found that the data was better described by a model in which variance increased as the square of the number of inputs. In other words, the standard deviation scales linearly with the number of inputs, similar to “scalar variability” models used for numerical and time interval estimation. This result held both for freely moving rats and for head-restrained rats; the latter suggests the phenomenon can be investigated using cellular-resolution two-photon microscopy.

Short-term memory is an essential higher cognitive function, allowing us to mentally store information, and thereby separate sensory input from motor output. Guided by previous neural recordings conducted in the rat FOF by Jeff Erlich et al. 2011 during a simple memory guided orienting task, we sought to test the hypothesis that activity in the rat FOF is causally maintaining the rat’s choice for its upcoming motor act. Surprisingly, as information encoded in the FOF increased, the ability to perturb the rat’s choice by disrupting that information decreased. We conducted the same experiments on the deep motor layers of the superior colliculus (SC) and found nearly identical results.   In order to reconcile these seemingly contradictory optogenetic and electrophysiological results we adapted a simple mutually inhibiting dynamical attractor network model. At the start of a trial the network resides on the unstable hill separating two basins of attraction, at this point there is little information encoded, but unilateral inactivation of either the FOF or SC can easily push the network off the hill into one of the attractors, biasing the decision. Later, during the memory delay period, when the network resides in one of the attractors, there is a large amount of information encoded, but now the same perturbation is insufficient to push the network out of the attractor. These results provide the first direct causal test that short-term memory is maintained by a network distributed across cortical and subcortical regions and is well characterized by attractor dynamics.

Executive control, defined by flexible sensorimotor remapping in response to changing environmental demands, is a remarkable feature of adaptive behavior, and is predominantly studied in primates and putatively mediated by the prefrontal cortex (PFC). To study executive control in the rat, graduating student Ann Duan, together with Jeff and Carlos, developed a novel behavior in which subjects are cued, on each trial, to apply a sensorimotor association to orient either toward a visual target (“Pro”) or away from it (“Anti”). Multiple behavioral asymmetries suggest that Anti behavior is cognitively demanding while Pro is easier to learn and perform. This is consistent with a prominent hypothesis in the primate literature that Anti requires prefrontal cortex (PFC), whereas Pro could be mediated by midbrain superior colliculus (SC). First author Ann Duan directly tested this hypothesis via reversible inactivation of the PFC or SC during task performance. Pharmacological inactivation of rat medial PFC supported its expected role in Anti. Remarkably, bilateral SC inactivation substantially impaired Anti while leaving Pro essentially intact. Moreover, SC inactivation eliminated the performance cost of switching from Anti to Pro tasks. Our results establish a rodent model of single-trial sensorimotor remapping and suggest a critical role for SC in the cognitively demanding Anti task.

Numerous brain regions have been shown to have neural correlates of gradually accumulating evidence for decision-making, but the causal roles of these regions in decisions driven by accumulation of evidence have yet to be determined. Here, in rats performing an auditory evidence accumulation task, we inactivated the frontal orienting fields (FOF) and posterior parietal cortex (PPC), two rat cortical regions that have neural correlates of accumulating evidence and that have been proposed as central to decision-making. We used a detailed model of the decision process to analyze the effect of inactivations. Inactivation of the FOF induced substantial performance impairments that were quantitatively best described as an impairment in the output pathway of an evidence accumulator with a long integration time constant (>240 ms). In contrast, we found a minimal role for PPC in decisions guided by accumulating auditory evidence, even while finding a strong role for PPC in internally-guided decisions.

Tim Hanks and Chuck Kopec took our temporal accumulation of auditory evidence decision task (“Poisson Clicks”, Brunton et al. 2013), and used it to investigate encoding of accumulating evidence in posterior parietal cortex (PPC) and frontal orienting fields (FOF) of rats. The paper develops a new method to estimate tuning curves for accumulating evidence, allowing us to plot firing rate as a function of value of the accumulated evidence. This method revealed that while PPC encoded accumulating evidence in a graded manner, the FOF did so in a more binary-like categorical manner, more consistent with encoding response selection (“if the GO signal came now, which of the two options does the accumulated evidence tell me I should choose?”) than with encoding the gradually accumulating evidence itself.

Information about response selection needs to be read out, and will affect behavior, only around the time of the GO signal. Such an encoding thus predicts that perturbing the FOF should have an effect only if the perturbation occurs at the time of the GO signal. High time-resolution optogenetic inactivation of the FOF confirmed this prediction, supporting the idea that the FOF is part of transforming accumulated evidence into a categorical choice, but is not part of the gradually accumulating evidence process itself.

Ben Scott, a postdoc in the Brody and Tank labs, developed a system for voluntary head-fixation in rats. Based on the principles of kinematic mounts often used in optics, the system allows precise re-positioning of a rat’s head, across multiple trials of a behavior, with an accuracy of a few microns. This enables cellular-resolution calcium imaging in behaving rats. In addition, because the rats engage the system voluntarily, the approach is amenable to high-throughput training. Thanks to Ben’s work, we can now use the lab’s training facility to teach voluntarily head-fixing rats to perform complex cognitive behaviors requiring many months to train. We are using this is to perform the first cellular-resolution imaging assays of neural activity involved in higher cognitive processes.

Where do imperfections in decision-making come from? And can rats perform gradual accumulation of evidence for decision-making, like primates do? This paper describes  a decision-making task that is particularly well-suited to detailed quantitative modeling and analysis. Using this task together with a trial-by-trial model of the behavior, first-author Bing Brunton showed that rats, like humans, can gradually accumulate evidence for decision-making, and that their evidence accumulator is optimal in the sense of being noiseless and lossless. Thus imperfections in decision-making all came from imperfections in sensory processing, none from the evidence accumulation process.The modeling and analysis methods Bing developed are the most statistically powerful approach to date for characterizing properties of decision-making processes. The approach can be readily applied to other decision-making tasks and modalities, and provides a moment-by-moment, trial-by-trial estimate of the subject’s internal estimate of the accumulating evidence.

Nature published a News and Views piece on our paper.

 

.

2011

 

Do rats have a brain region homologous to the primate Frontal Eye Fields (FEF)? We trained rats to perform memory-guided orienting, and used that to investigate the functional role of the rat’s Frontal Orienting Fields (FOF). Memory-guided orienting is a classic task in which a transient sensory cue tells you in which direction you should orient, but you must withhold your motion over a delay period, until a “go” cue arrives. (You must therefore keep either the sensory cue or your intended motion in memory during the delay period.) First-author Jeff Erlich recorded in the Frontal Orienting Fields (FOF), a rat cortical area that has been suggested as homologous to the primate FEF. Just as in the primate FEF, he found persistent neural activity during the delay period that predicts the subject’s subsequent orienting motion. Again as in the primate FEF, Jeff found that pharmacological inactivation of the FOF impairs memory-guided orienting markedly more than it impairs sensory-guided orienting. Our data support the rat FOF / primate FEF homology.  Full Description.

Which brain areas are involved in time discrimination? Using the same behavioral protocol and stimuli, rats were trained to either discriminate high versus low tones, or discriminate long versus short durations. First-author Shraddha Pai then lesioned either the auditory cortex (ACx) or the medial prefontal cortex (mPFC). ACx lesions led to a clear impairment in frequency discrimination, but no impairment in the matched duration discrimination task. mPFC lesions led to no detectable impairment in either task. The data suggests that time discrimination may bypass both ACx and mPFC.

Quantifying the location and/or number of features in a histological section of the brain currently requires one to first, manually register a corresponding section from a tissue atlas onto the experimental section and second, count the features. No automated method exists for the first process (registering), and most automated methods for the second process (feature counting) operate reliably only in a high signal-to-noise regime. To reduce experimenter bias and inconsistencies and increase the speed of these analyses, we developed Atlas Fitter, a semi-automated, open-source MatLab-based software package that assists in rapidly registering atlas panels onto histological sections. We also developed CellCounter, a novel fully automated cell counting algorithm that is designed to operate on images with non-uniform background intensities and low signal-to-noise ratios.

 

.

2009 – 2010

How do populations of neurons multiplex different types of information? Recordings in prefrontal cortices typically show neurons that have responses that depend on multiple aspects of the task being performed, and that are highly heterogeneous: although each neuron is reliable, no two neurons are alike. Faced with datasets of recordings of hundreds of such neurons, people often first use principal components analysis (PCA), so as to look at a reduced number of dimensions that captures most of the variance. But PCA doesn’t orient its axes in order to clearly distinguish task components. In this paper, we describe a method to automatically find the orientation of the PC axes that best distinguish between the components of the task we’re interested in. This allows a particularly clear and intelligible view of the data.


Older papers

[this section on older papers needs filling in]

  • Machens, C.K., R. Romo, and C.D. Brody. 2005. Flexible control of mutual inhibition: a neural model of two-interval discrimination. Science 307: 1121–1124.
  • Hopfield, J.J., and C.D. Brody. 2004. Learning rules and network repair in spike-timing-based computation networks. Proc. Natl. Acad. Sci. USA 101: 337–342.
  • Brody, C.D., and J.J. Hopfield. 2003. Simple networks for spike-timing-based computation, with application to olfactory processing. Neuron 37: 843–852.
  • Hopfield, J.J., and C.D. Brody. 2001. What is a moment? Transient synchrony as a collective mechanism for spatiotemporal integration. Proc. Natl. Acad. Sci. USA 98: 1282–1287.
  • Romo, R., C.D. Brody, A. Hernandez, and L. Lemus. 1999. Neuronal correlates of parametric working memory in the prefrontal cortex. Nature 399: 470–473.