Single-neuron activity in prefrontal cortex (PFC) is tuned to mixtures of multiple task-related aspects. Such mixed selectivity is highly heterogeneous, seemingly disordered and therefore difficult to interpret. We analysed the neural activity recorded in monkeys during an object sequence memory task to identify a role of mixed selectivity in subserving the cognitive functions ascribed to PFC. We show that mixed selectivity neurons encode distributed information about all task-relevant aspects. Each aspect can be decoded from the population of neurons even when single-cell selectivity to that aspect is eliminated. Moreover, mixed selectivity offers a significant computational advantage over specialized responses in terms of the repertoire of input-output functions implementable by readout neurons. This advantage originates from the highly diverse non-linear selectivity to mixtures of task-relevant variables, a signature of high-dimensional neural representations. Crucially, this dimensionality is predictive of animal behaviour as it collapses in error trials. Our findings suggest to move the focus of attention from neurons that exhibit easily interpretable response tuning to the widely observed, but rarely analysed, mixed selectivity neurons.

Major depression is characterized by diverse debilitating symptoms that include hopelessness and anhedonia1. Dopamine neurons involved in reward and motivation2–9 are among many neural populations that have been hypothesized to be relevant10, and certain antidepressant treatments, including medications and brain stimulation therapies, can influence the complex dopamine system. Until now it has not been possible to test this hypothesis directly, even in animal models, as existing therapeutic interventions are unable to specifically target dopamine neurons. Here we investigated directly the causal contributions of defined dopamine neurons to multidimensional depression-like phenotypes induced by chronic mild stress, by integrating behavioural, pharmacological, optogenetic and electrophysiological methods in freely moving rodents. We found that bidirectional control (inhibition or excitation) of specified midbrain dopamine neurons immediately and bidirectionally modulates (induces or relieves) multiple independent depression symptoms caused by chronic stress. By probing the circuit implementation of these effects, we observed that optogenetic recruitment of these dopamine neurons potently alters the neural encoding of depression-related behaviours in the downstream nucleus accumbens of freely moving rodents, suggesting that processes affecting depression symptoms may involve alterations in the neural encoding of action in limbic circuitry.

Behavioural states in mammals, such as the anxious state, are characterized by several features that are coordinately regulated by diverse nervous system outputs, ranging from behavioural choice patterns to changes in physiology (in anxiety, exemplified respectively by risk-avoidance and respiratory rate alterations). Here we investigate if and how defined neural projections arising from a single coordinating brain region in mice could mediate diverse features of anxiety. Integrating behavioural assays, in vivo and in vitro electrophysiology, respiratory physiology and optogenetics, we identify a surprising new role for the bed nucleus of the stria terminalis (BNST) in the coordinated modulation of diverse anxiety features. First, two BNST subregions were unexpectedly found to exert opposite effects on the anxious state: oval BNST activity promoted several independent anxious state features, whereas anterodorsal BNST-associated activity exerted anxiolytic influence for the same features. Notably, we found that three distinct anterodorsal BNST efferent projections-to the lateral hypothalamus, parabrachial nucleus and ventral tegmental area-each implemented an independent feature of anxiolysis: reduced risk-avoidance, reduced respiratory rate, and increased positive valence, respectively. Furthermore, selective inhibition of corresponding circuit elements in freely moving mice showed opposing behavioural effects compared with excitation, and in vivo recordings during free behaviour showed native spiking patterns in anterodorsal BNST neurons that differentiated safe and anxiogenic environments. These results demonstrate that distinct BNST subregions exert opposite effects in modulating anxiety, establish separable anxiolytic roles for different anterodorsal BNST projections, and illustrate circuit mechanisms underlying selection of features for the assembly of the anxious state.

The prefrontal cortex (PFC) is thought to participate in high-level control over generation of behaviors (including the decision to execute actions1); indeed, imaging and lesion studies in human beings have revealed that PFC dysfunction can lead to either impulsive states with increased tendency to initiate action2, or to amotivational states characterized by hopeless and depressive symptoms3. In light of the opposite valence of these two phenotypes as well as the broad complexity of other tasks attributed to PFC, it may be that only a specific subclass of PFC neurons favors effortful behavioral responses to challenging situations. Here we develop and employ a quantitative method for continuous assessment and control of active response to behavioral challenge, synchronized with single-unit electrophysiology and optogenetics in freely moving rats. We first observed by recording in the medial PFC (mPFC) that many neurons were not simply movement-related in their spike firing patterns, but instead were selectively modulated from moment to moment according to the animal’s decision to act in a challenging situation. Surprisingly, we next found that direct activation of principal neurons in the mPFC had no detectable causal effect on this behavior. We next tested if this behavior could be causally mediated by only a subclass of mPFC cells defined by specific downstream wiring. Indeed, by leveraging optogenetic projection-targeting to control cells with specific efferent wiring patterns, we observed that selective activation of those mPFC cells projecting to the brainstem dorsal raphe nucleus (DRN), a serotonergic nucleus implicated in major depressive disorder4, induced a profound, rapid, and reversible effect on selection of the active behavioral state. These results may be of importance in understanding the circuit basis of normal and pathological patterns of action selection and motivation in behavior.