Researchers at MIT have mapped a previously underappreciated brain circuit that alters how visual information is processed depending on an animal’s level of arousal and movement. In a study in mice, they show that the prefrontal cortex sends specialized feedback signals to visual and motor regions, tightening or loosening visual representations according to behavioral state, as reported in Neuron.
Vision is not just a passive recording of the outside world but is actively shaped by internal states, according to new research from MIT.
In work summarized by the Picower Institute at MIT, the study reports that the prefrontal cortex — a major hub for executive control — sends tailored signals to the primary visual cortex (VISp) and the primary motor cortex (MOp) in mice.
These signals adjust how those regions operate based on factors such as how alert the animal is and whether it is moving, either sharpening or dampening visual details to prioritize relevant information.
The team, led by postdoctoral researcher Sofie Ährlund-Richter, with senior author Mriganka Sur, the Paul and Lilah Newton Professor in The Picower Institute for Learning and Memory and MIT's Department of Brain and Cognitive Sciences, focused on two prefrontal subregions: the orbitofrontal cortex (ORB) and the anterior cingulate area (ACA).
According to the ScienceDaily report on the work, both ORB and ACA relay information about arousal and movement to VISp and MOp, but their effects differ. Higher arousal increased ACA’s tendency to help VISp sharpen visual representations, particularly for more uncertain or harder-to-detect stimuli. ORB became influential only when arousal was very high, and its involvement appeared to reduce the clarity of visual encoding, potentially dampening responses to strong but less relevant stimuli.
"These two PFC subregions are kind of balancing each other," Ährlund-Richter said. "While one will enhance stimuli that might be more uncertain or more difficult to detect, the other one kind of dampens strong stimuli that might be irrelevant."
To map these pathways, Ährlund-Richter carried out detailed anatomical tracing of the connections from ACA and ORB to VISp and MOp. In additional experiments, mice ran on a wheel while viewing structured images or naturalistic movies at different contrast levels, and brief air puffs were used to increase arousal.
Throughout these tasks, the researchers recorded neural activity in ACA, ORB, VISp and MOp, with particular attention to signals traveling along the axons linking prefrontal and posterior areas. The tracing work showed that ACA and ORB each communicate with multiple cell types in their target regions and follow distinct spatial patterns: in VISp, ACA primarily targeted layer 6, while ORB communicated mainly with neurons in layer 5.
When the team examined the transmitted information and neural activity, several patterns emerged. ACA neurons conveyed more detailed visual information than ORB neurons and were more responsive to changes in contrast. ACA activity also closely tracked arousal level, whereas ORB responded only when arousal reached a relatively high threshold. When signaling to MOp, both prefrontal regions carried information about running speed; when signaling to VISp, they signaled whether the mouse was moving or still, and they also carried arousal-related signals and a small amount of visual detail to MOp.
To test how this communication shapes visual processing, the scientists temporarily blocked feedback pathways from ACA and ORB to VISp and then measured how VISp neurons responded in the absence of those inputs. They found that ACA and ORB exerted specific and opposing influences on visual encoding that depended on the mouse’s movement and level of arousal.
"That's the major conclusion of this paper: There are targeted projections for targeted impact," Sur said in the Picower Institute account of the findings. The authors write that their data support a model of prefrontal feedback that is specialized at the level of both prefrontal subregions and their targets, allowing each region to selectively shape target-specific cortical activity rather than modulating it globally.
In addition to Ährlund-Richter and Sur, the research team included Yuma Osako, Kyle R. Jenks, Emma Odom, Haoyang Huang and Don B. Arnold. According to the ScienceDaily summary, the work was supported by a Wenner-Gren foundations Postdoctoral Fellowship, the National Institutes of Health and the Freedom Together Foundation.
The study, titled "Distinct roles of prefrontal subregion feedback to the primary visual cortex across behavioral states," was published in Neuron in 2025.