Douglas A. Ruff

Neurons throughout the brain encode essentially everything, so why have distinct areas? We proposed that areas format different information sources differently, implying their computational role. Here we compared the middle temporal (MT) area, involved in visual motion perception, and the dorsolateral prefrontal cortex (dlPFC), implicated in decision-making and reward. In a task that required monkeys to combine motion and reward information, both regions encoded both motion and reward, but formatted differently. Area MT represented them separably, whereas the dlPFC combined them in a decision-related format. A recurrent network model trained to reproduce that formatting predicted a distinct effect of stimulating those areas. Consistent with predictions, MT stimulation biased choices whereas dlPFC stimulation produced 'winner-take-all' outcomes. Our findings suggest that a modular structure flexibly reformats information to enable complex behaviors.

Alzheimer’s disease and related dementias are typically described at two levels: the accumulation of molecular pathology and the emergence of cognitive impairment. Understanding the relationship between pathology, often studied in animal models, and human cognition will require measurements spanning intermediate scales, including single neurons, neuronal populations, and distributed networks. Here we combine longitudinal measurements of behavior and neuronal population activity with fluid and histological biomarkers in a macaque model of early-stage disease. We find in two animals that visually guided behavior becomes increasingly disorganized, with less consistent and more variable patterns of exploration, despite preserved performance on simple tasks. In parallel, coordinated activity within and between neuronal populations in visual and parietal cortex declines, even as single-neuron tuning and basic feature encoding remain stable. The magnitude of these physiological changes was broadly consistent with biomarker progression. These changes arise when pathology is largely confined to regions providing feedback to visual cortex, indicating that functional disruption extends beyond sites of prominent pathology. Together, these results show that early disease progression is not marked by the loss of individual functions at any single level, but by a selective disruption of coordination across levels, from neuronal populations to behavior. This disorganized state is measurable and modifiable: methylphenidate administration was associated with a transient restoration of behavioral organization. These findings identify disruption of neuronal population organization as a defining feature of early-stage Alzheimer’s disease and establish coordinated population activity as a candidate target for therapeutic intervention.

Intelligent behavior depends on the brain's ability to represent multiple features of the environment simultaneously while keeping those representations independent [1,2]. Patients with Alzheimer's disease often mix up objects, people, and events [3-7], raising the possibility that disease mixes up the way that information is represented in the brain. Here we show that the independence of visual representations progressively breaks down during early stages of disease progression in a rhesus macaque model of Alzheimer's disease and related dementias [8-10]. In visual area V4, representations of different visual features become progressively less independent, such that the representation of one feature is increasingly influenced by the value of another. We term this loss of independence neuronal feature confusion. This neuronal change predicts a specific behavioral consequence: because feature representations become less independent, preferences associated with one visual feature increasingly influence visually guided choices associated with other, independent features. Using an analogous image-selection task, we found the same behavioral signature in people with mild cognitive impairment, distinguishing them from age-matched controls. These results identify a specific and measurable alteration in neuronal population representations that predicts a behavioral change observed across species. More broadly, these findings demonstrate that neuronal population representations can guide the development of sensitive, non-invasive behavioral methods for early detection of functional changes associated with Alzheimer's disease.

Alzheimer’s disease (AD) and related dementias (ADRD) are defined by molecular and cellular pathology and cognitive decline, but linking these levels requires understanding how pathology alters large-scale neuronal activity. We longitudinally tracked behavior, multi-area neuronal population activity, and fluid and histological biomarkers in a macaque model of early-stage ADRD. As pathology progressed, visually guided behavior became increasingly disorganized, reflected in less structured exploration despite preserved task performance. Guided by systems neuroscience principles linking neuronal population activity with organized goal-directed behavior, we found progressive reductions in coordinated neuronal population activity within and between visual and parietal cortices, even as single-neuron tuning and basic feature encoding remained stable. These changes emerged when tau pathology was largely confined to regions providing feedback to visual cortex. This disorganized state appears modifiable: proof-of-concept methylphenidate administration was associated with transient improvement in behavioral organization. Together, these findings identify disruption of neuronal population organization as a defining feature of early-stage ADRD and frame early dysfunction as a disorder of coordinated population activity.

Neurons in higher-order visual areas integrate information through a canonical computation called normalization. The strength of normalization is highly heterogeneous across neurons, and this heterogeneity correlates with attention-mediated modulations in neural responses. However, the circuit mechanism underlying the heterogeneous normalization strength is unclear. In this work, we study normalization in a spiking neuron network model of visual cortex. Our model reveals that the heterogeneity of normalization strength is highly correlated with the inhibitory current each neuron receives. The correlation between inhibition and other synaptic inputs explains the experimentally observed dependence of spike count correlations on normalization strength. Further, we find that neurons with stronger normalization encode information more efficiently, and that networks with more heterogeneity in normalization encode visual stimuli with higher information and capacity. Together, our model provides a mechanistic explanation of heterogeneous normalization strengths in the visual cortex and sheds light on the computational benefits of neuronal heterogeneity.

Effective decision-making requires integrating multiple information sources, weighted by their reliability and context. While classic studies show near-optimal cue combination in extensively controlled laboratory settings and lengthy inperson experiments, everyday choices often occur in less controlled environments. We examined cue combination under these conditions using an online perceptual estimation task in large and diverse participant cohorts. Participants combined cues, including visual motion direction, spatial visual information, and sound. We investigated cue combination with and without cue conflict. Performance varied with age and self-reported ADHD or Autism. Visual cues were combined more optimally than audio-visual combinations. We observed qualitative similarities in an analagous task with non-human primates. We used electrical microstimulation in non-human primates, targeting unimodal or cross-modal association areas. Stimulation of visual cortex was integrated with sensory motion cues, while stimulation of prefrontal cortex promoted winner-take-all choices. These findings suggest distinctions between within- and across-modality integration, with deviations potentially informative of age or neurodiversity.

Shared trial-to-trial variability across sensory neurons is reliably reduced when perceptual performance improves, yet this variability is low-dimensional, so it could be ignored by an optimal readout mechanism. Why then is it so consistently related to behavior? We propose that shared variability both reflects circuit structure and reveals the information communicated to downstream areas . In this framework, the same connectivity that shapes signal propagation also shapes shared variability. Using a circuit model, we show that when sensory signals align with shared variability, behaviorally relevant information is amplified without compromising coding fidelity. Analyses of neural population recordings from multiple brain areas and tasks reveal that the dominant axis of shared variability consistently aligns with task-relevant stimulus features and action plans . Finally, the behavioral impact of microstimulation can be explained by the extent to which it changes projections onto the shared variability axis. These findings suggest that shared variability may illuminate, rather than obscure, the neural dimensions that guide behavior. Significance Statement: The brain’s ability to use different features of sensory information flexibly across many tasks is essential for complex decision-making. Our study reveals that the correlated variability in neural responses in mid-level visual areas indicates the visual feature that is relevant for behavior. Our biologically plausible network model predicted that it is beneficial for the representation of behaviorally relevant visual information to be aligned with the correlated variability in neurons. We tested and confirmed this in five independent neural data sets. Our results suggest that trial-by-trial variability does not affect the information encoded in sensory neurons but instead is a valuable signal for us to understand which combination of the encoded features is being used by the brain to guide choices.

Animals capable of complex behaviors tend to have more distinct brain areas than simpler organisms, and artificial networks that perform many tasks tend to self-organize into modules (1-3). This suggests that different brain areas serve distinct functions supporting complex behavior. However, a common observation is that essentially anything that an animal senses, knows, or does can be decoded from neural activity in any brain area (4-6). If everything is everywhere, why have distinct areas? Here we show that the function of a brain area is more related to how different types of information are combined (formatted) in neural representations than merely whether that information is present. We compared two brain areas: the middle temporal area (MT), which is important for visual motion perception (7, 8), and the dorsolateral prefrontal cortex (dlPFC), which is linked to decision-making and reward expectation (9,10)). When monkeys based decisions on a combination of motion and reward information, both types of information were present in both areas. However, they were formatted differently: in MT, they were encoded separably, while in dlPFC, they were represented jointly in ways that reflected the monkeys' decision-making. A recurrent neural network (RNN) model that mirrored the information formatting in MT and dlPFC predicted that manipulating activity in these areas would differently affect decision-making. Consistent with model predictions, electrically stimulating MT biased choices midway between the visual motion stimulus and the preferred direction of the stimulated units (11), while stimulating dlPFC produced 'winner-take-all' decisions that sometimes reflected the visual motion stimulus and sometimes reflected the preference of the stimulated units, but never in between. These results are consistent with the tantalizing possibility that a modular structure enables complex behavior by flexibly reformatting information to accomplish behavioral goals.

How neural representations preserve information about multiple stimuli is mysterious. Because tuning of individual neurons is coarse (for example, visual receptive field diameters can exceed perceptual resolution), the populations of neurons potentially responsive to each individual stimulus can overlap, raising the question of how information about each item might be segregated and preserved in the population. We recently reported evidence for a potential solution to this problem: when two stimuli were present, some neurons in the macaque visual cortical areas V1 and V4 exhibited fluctuating firing patterns, as if they responded to only one individual stimulus at a time. However, whether such an information encoding strategy is ubiquitous in the visual pathway and thus could constitute a general phenomenon remains unknown. Here we provide new evidence that such fluctuating activity is also evoked by multiple stimuli in visual areas responsible for processing visual motion (middle temporal visual area, MT), and faces (middle fundus and anterolateral face patches in inferotemporal cortex – areas MF and AL), thus extending the scope of circumstances in which fluctuating activity is observed. Furthermore, consistent with our previous results in the early visual area V1, MT exhibits fluctuations between the representations of two stimuli when these form distinguishable objects but not when they fuse into one perceived object, suggesting that fluctuating activity patterns may underlie visual object formation. Taken together, these findings point toward an updated model of how the brain preserves sensory information about multiple stimuli for subsequent processing and behavioral action. Impact Statement: We find neural fluctuations in multiple areas along the visual cortical hierarchy that could allow the brain to represent distinct co-occurring visual stimuli.

Significance: Stimulants such as methylphenidate (Ritalin) are often used clinically to treat attention deficit hyperactivity disorder (ADHD) and generally to improve selective attention. Their systemic administration provides a causal manipulation of the attentional system, allowing measurements of the neuronal population changes that accompany artificially manipulated selective attention. We found that orally administering methylphenidate to rhesus monkeys selectively improved visual performance at only their attended spatial location, and that correspondingly specific changes in the shared variability of groups of neurons in visual area V4 occurred only when attention was directed toward the receptive fields of those neurons. Critically, methylphenidate changed behavior exactly when it changed the shared variability of the neuronal responses, suggesting that it may work through naturally selective cognitive mechanisms.