Our interaction with the environment hinges on extracting meaningful signals that signal potential threats or rewards. The dmPFC, a brain region pivotal in higher cognitive functions, occupies a central role in translating these sensory inputs into motivated actions. Despite its acknowledged importance, how this area processes complex cue information—specifically how it segregates and integrates salience (importance), valence (positive or negative emotional charge), and value (expected outcome magnitude)—has remained a mystery until now. By harnessing calcium imaging technology in freely behaving male mice, Winke, Lüthi, Herry, and colleagues achieved an unprecedented granularity in observing single-neuron activity as the animals learned to discriminate stimuli predicting varying outcomes of reward or punishment.

The methodology leveraged in this study represents a major advancement. Using calcium indicators that fluoresce upon neuronal activation, the researchers tracked population-level dmPFC neural dynamics longitudinally across associative learning phases. This approach permitted real-time, cellular-resolution mapping of how neurons encode stimulus features relevant to motivational salience and affective value. By presenting mice with cues linked to either positive (appetitive) or negative (aversive) outcomes, the study created a framework enabling the dissociation of neural representations for salience, valence, and value—dimensions often conflated in previous investigations.

One of the study’s central findings concerns the geometrical organization of neuronal activity within the dmPFC. Neuronal populations were found to encode appetitive and aversive values along distinct, orthogonal axes—essentially in separate neural “spaces.” This segregation allows simultaneous but independent processing of positive and negative motivational aspects of stimuli, preventing ambiguity in the organism’s behavioral response. Notably, subsets of neurons within the dmPFC were specialized to encode either valence or the stimulus’ salience, highlighting a multifaceted coding strategy that supports adaptive decision-making. Such orthogonality in neural representation represents a novel architectural principle for the neural basis of motivated behaviors.

Equally revealing was how these neural representations evolved dynamically during learning. Initial generalized responses to environmental cues sharpened over sessions, with specific dmPFC populations emerging to distinctly encode the learned value associated with each stimulus. This plasticity suggests that the dmPFC does not merely relay sensory information but actively configures its networks to optimize behavior by encoding motivational parameters in well-defined formats. This learning-related neural geometry ensures that the brain’s motivational circuits can flexibly adjust focus between reward-seeking and threat avoidance, based on contextual cue significance.

Beyond basic neuroscience, these findings have profound implications for conceptualizing the neural underpinnings of complex cognitive functions. Disorders such as anxiety, addiction, and depression often involve maladaptive valuations of environmental stimuli—where either threatening signals are exaggerated or rewarding cues are undervalued. Understanding how the dmPFC orchestrates the representation of motivational dimensions opens new avenues for targeted interventions that could recalibrate disordered neural coding patterns. The study’s insights into the geometry of neural representations provide a platform for developing biomimetic models that simulate decision-making disruptions observed in psychiatric illnesses.

From a broader evolutionary perspective, the demonstrated population coding mechanisms align with theories that prefrontal cortex networks evolved to handle complex, multidimensional valuation computations essential for survival. Encoding valence, salience, and value along distinct neural channels may confer computational advantages, allowing organisms to rapidly prioritize behaviors when confronted with competing stimuli in unpredictable environments. The exquisite tuning of dmPFC neurons to motivationally salient cues highlights the sophistication of mammalian cognition, bridging sensorimotor integration with affective processing.

Importantly, the researchers emphasize that the multifunctional nature of the dmPFC neural ensembles enables concurrent processing of opposing motivational forces rather than a simplistic linear summation. This nuanced balance allows animals to dynamically bias their behavior towards either approach or avoidance depending on the perceived cost-benefit scenario. Consequently, the study’s population-level perspective contrasts with earlier paradigms that focused mainly on single-neuron responses, providing a more accurate depiction of the emergent properties underpinning motivated behavior.

Technological advancements in calcium imaging played a critical role here, as traditional electrophysiological recordings could not feasibly resolve the high-dimensional activity patterns across broad dmPFC populations in freely moving subjects. The researchers’ innovative integration of behavioral paradigms with in vivo imaging permitted a granular yet ecologically valid examination of associative learning processes, devoid of confounds introduced by restraint or anesthesia. This methodological synergy sets a new benchmark for investigating neural computations underlying complex behaviors.

The research team also suggests that these findings may extend to understanding human prefrontal cortex functionality. Given the conserved architecture and evolutionary continuity of prefrontal regions across mammals, insights gained from murine models offer translational value. Dissecting how valence and value are represented geometrically in human prefrontal networks could improve cognitive-behavioral approaches that depend on modifying incentive processing, such as therapies for addiction or mood disorders.

In conclusion, this pioneering study elucidates how the dmPFC encodes learned environmental cues in a high-dimensional neural space structured by the orthogonal encoding of salience, valence, and value. Such dynamic neural geometry governs the delicate balance of motivated behavior, enabling organisms to efficiently navigate complex environments by judiciously weighing opportunities and threats. The findings redefine our understanding of prefrontal cortex functionality and establish a framework for future research aimed at unraveling the neural algorithms behind decision-making and emotional regulation.

As neuroscience continues to unravel the neural language of behavior, this research represents a milestone demonstrating the intricate cortical choreography that underlies even the simplest decisions. By capturing the population-level dynamics of neuronal ensembles, the authors illuminate principles that could inspire artificial intelligence models seeking to emulate adaptive, value-based behaviors. Ultimately, this work exemplifies how advanced neurotechnologies can deepen our grasp of brain function and its role in shaping motivated actions fundamental to life itself.

Subject of Research: Neural Encoding of Motivational Cue Dimensions in the Dorsomedial Prefrontal Cortex

Article Title: Prefrontal neural geometry of learned cues guides motivated behaviours

Article References:
Winke, N., Lüthi, A., Herry, C. et al. Prefrontal neural geometry of learned cues guides motivated behaviours. Nature (2026). https://doi.org/10.1038/s41586-025-09902-2

Image Credits: AI Generated

DOI: https://doi.org/10.1038/s41586-025-09902-2

Keywords: dorsomedial prefrontal cortex, dmPFC, motivated behaviour, salience, valence, value coding, calcium imaging, neural population dynamics, associative learning, neural geometry, reward, punishment, decision-making

At the crux of this investigation lies the motor cortex—the brain’s command center orchestrating intentional movement. While prior research predominantly focused on sensory cortices and stimulus-driven responses, this study shifts the spotlight onto motor cortex neurons as mice engage in multidirectional tongue-reaching tasks to retrieve rewards. This ethologically relevant behavior occurs without prior conditioning, thereby capturing neural dynamics in their unperturbed, innate state.

The researchers meticulously charted neural activity patterns from layer 2/3 of the motor cortex, revealing that individual neurons exhibit selective tuning both to the spatial goals of movements and the outcomes of rewarded actions. These neural signals are not confined to the moments during movement itself but also extend into periods immediately following the action, suggesting sustained representations of behaviorally salient information.

To unravel the microcircuit architecture driving these activity patterns, the team leveraged powerful all-optical methods that allowed causal elucidation of effective neural connections with unparalleled precision. By simultaneously stimulating and recording from a prodigious number of excitatory neurons—more than twenty million pairwise connections evaluated—they painted a comprehensive map of synaptic influences shaping motor cortex processing.

A defining feature emerging from this connectomic landscape is a multi-scale, columnar organization within the excitatory networks. Neurons preferentially formed local “like-to-like” excitatory connections within a fine spatial scale under 100 microns, grouped according to their directional tuning to movement targets. Complementing this local excitation was a broader pattern of spatially extended inhibition, establishing a balance ramped across cortical space.

Interestingly, the connectivity did not conform to a simplistic all-or-nothing pattern but rather exhibited a continuous spectrum. Most neurons were sparsely connected, engaging with relatively few partners, whereas a rare subset functioned as densely connected hubs with extensive influence over their neighbors. These hub neurons exhibited a paradoxical weak tuning to specific movement targets and reward outcomes yet wielded outsized network control, likely playing a modulatory role.

The implications of these findings are profound. The interplay of finely tuned local excitatory circuits with wider inhibitory regulation and the presence of hub nodes suggests a versatile substrate capable of rapidly encoding and updating motor goals. This architecture may be evolutionarily optimized to enable swift learning and flexible adjustment of complex, goal-directed behaviors pivotal for survival.

Beyond deepening our grasp of motor cortex microcircuitry, this work establishes a technological roadmap for probing causal connectivity at scale in vivo. The all-optical approach, combining precise neuronal perturbations with high-resolution functional imaging, surmounts longstanding technical obstacles that previously limited mechanistic insight into large-scale neural networks underpinning behavior.

Moreover, since the study examines naturalistic behavior absent of training, it captures the foundational wiring and coding frameworks from which learning and plasticity can emerge. Dissecting these primal networks lays essential groundwork not only for neuroscience but also for neuromorphic computing and the development of brain-inspired robotic control systems.

The discovery that neural hubs—despite weak sensory tuning—assert broad control points to a sophisticated organizational principle. Such nodes may coordinate widespread cortical dynamics, integrating diverse inputs and orchestrating coherent motor commands critical for achieving multiple goals under varying contexts.

This comprehensive map of motor cortex connectivity harmonizes with longstanding theories positing that efficient movement control relies on balanced excitation and inhibition sculpted at multiple spatial and temporal scales. By detailing how these forces manifest at the cellular and microcircuit level, the study bridges scales from molecules to behavior.

Crucially, this research underscores the importance of studying cortical circuits in ethologically valid settings. Artificial training protocols may obscure intrinsic wiring motifs or introduce plasticity artifacts absent in natural behavior. Thus, observing the brain’s default architecture offers a more faithful window into how neural ensembles encode and execute voluntary actions.

Taken together, these insights elevate our knowledge of the neural basis of purposeful movement and pave avenues for future investigations into how hierarchical and distributed cortical networks collaborate during learning, decision-making, and adaptation. Such understanding holds promise not only for treating motor dysfunction but also for designing intelligent machines capable of fluid, adaptive behaviors.

As we stand on the cusp of a new era in systems neuroscience, this landmark study exemplifies how merging cutting-edge optical technologies with behavioral paradigms can unlock the deepest secrets of brain connectivity—secrets that ultimately define who we are as behaving, goal-driven beings.

Subject of Research: Motor cortex connectivity underlying goal-directed behavior in mice.

Article Title: Connectivity underlying motor cortex activity during goal-directed behaviour.

Article References:
Finkelstein, A., Daie, K., Rózsa, M. et al. Connectivity underlying motor cortex activity during goal-directed behaviour. Nature (2025). https://doi.org/10.1038/s41586-025-09758-6

DOI: https://doi.org/10.1038/s41586-025-09758-6