Tracking self-performance in the prefrontal cortex: It’s layered

Abstract

In this issue of Cell, Spellman and colleagues record and manipulate the activity of neurons in the medial prefrontal cortex of mice performing a task in which they must pay attention to different stimuli. They show that this brain region is important for monitoring the animals’ performance, and neurons that appear to contribute to behavior reside in deep cortical layers.

An impressive but often effortless aspect of higher cognition is cognitive flexibility, or the ability to adjust our behavior depending on context. Cognitive flexibility, which is impaired in neuropsychiatric and neurodegenerative disorders, is often assessed using attentional set-shifting tasks (Brown & Tait, 2016). These tasks involve attending to a relevant stimulus over an irrelevant one to enhance processing of the relevant stimulus. For example, on an airplane, one might attend to their own music and tune out an announcement from the pilot. But if the plane encounters turbulence, one may switch their attention to the pilot’s message. Similarly, in attentional set-shifting tasks, subjects are required to exhibit a response based on a relevant stimulus while ignoring an irrelevant one. Following a rule change, the subject must shift their response to depend on the previously irrelevant stimulus. Rule changes are often uncued and must be inferred by tracking performance: the response that was correct according to the previous rule is no longer rewarded.

An extensive literature implicates the prefrontal cortex, including the rodent medial prefrontal cortex (mPFC), in attentional set-shifting (Wimmer et al., 2015; Bissonette et al., 2013). A leading idea is that mPFC mediates attentional set-shifting by biasing sensorimotor processing and decision-making (Miller & Cohen, 2001). In this issue of Cell, Spellman et al. (2021) provide an alternative explanation and novel circuit-level mechanism for the role of mPFC in cognitive flexibility.

If mPFC mediates attentional set-shifting via sensorimotor biasing, prefrontal neurons may modulate the activity of neurons in sensory and motor areas to favor the appropriate stimulus-response mapping for the current rule. Alternatively, since subjects must infer rule switches based on outcome feedback, the prefrontal cortex could monitor performance to infer the rule. While these hypotheses are not mutually exclusive, each one makes a distinct prediction about when prefrontal activity should be important for behavior. According to the attentional biasing hypothesis, the mPFC would be important for behavior during presentation of the sensory stimulus and selection of the response. For performance monitoring, the mPFC would be important for behavior at the time of outcome feedback (Figure 1A).

Figure 1. How does mPFC contribute to attentional set-shifting?

(A) Simplified timeline of the attentional set-shifting task used by Spellman et al. (2021). mPFC activity was necessary during the time of outcome feedback rather than during the trial, suggesting mPFC is important for performance monitoring.

(B) Response heterogeneity of mPFC cells in the present study was related to the layer in which the cell resided rather than the cell’s projection target (efferent connectivity).

Spellman et al. (2021) trained head-fixed mice on an attentional set-shifting task in which they were presented with whisker and odor stimuli. Mice were trained to lick a left or right water spout based on a feature of the relevant stimulus; if they responded correctly according to the current rule, they received a water reward. Optogenetically disrupting neural activity at the time of outcome feedback, but not during the trial, impaired behavior. These experiments support a role for the mPFC in performance monitoring during attentional set-shifting, prompting a revision of previous models of prefrontal control of cognitive flexibility.

Sources of response heterogeneity in prefrontal circuits

Neurons in the prefrontal cortex exhibit highly diverse responses during behavior. This dizzying heterogeneity has been referred to as the “cortical zoo of responses”. A major question in the field is whether diverse cortical responses derive from the responses of different cell types. Addressing this question requires determining which characteristics of cells differentiate their responses, or what is the most useful way to define a cell type.

Recent studies have found that neurons that project to the same brain area exhibit similar neural responses during behavior (Hirokawa et al., 2019; Otis et al., 2017). These data suggest that patterns of efferent connectivity, or where neurons project to, may account for response heterogeneity in prefrontal circuits.

Spellman et al. (2021) used viral and optical methods to record from mPFC neurons projecting to the ventromedial striatum (VMS) or the mediodorsal thalamus (MDT); both projections have been implicated in cognitive flexibility (Marton et al., 2018; Nakayama et al., 2018). Surprisingly, this study showed that both cell types exhibited highly similar activity. In contrast, a strong predictor of a neuron’s response was the cortical layer in which it resided (Figure 1B).

A defining feature of the cortex, including the mPFC, is its layered structure, which constrains patterns of neural connectivity. Long-range axons arriving from different brain areas often terminate in particular cortical layers and so are biased to synapse with neural processes in those layers. While neurons exhibiting similar efferent connectivity often reside in the same layer, a single cortical layer contains neurons projecting to multiple targets.

The authors observed stronger encoding of response and outcome in deep cortical layers for both mPFC→VMS and mPFC→MDT neurons. This organization could reflect patterns of afferent connectivity, or where synaptic inputs come from. To test this hypothesis, Spellman et al. (2021) used viral approaches to identify brain areas that differentially projected to the superficial and deep layers of mPFC. They found that the anterior cingulate cortex (ACC), which has been implicated in performance monitoring in other studies (Bissonette et al., 2013; Hyman et al., 2017), projects more strongly to deep layers of mPFC.

This raises the intriguing hypothesis that representations of responses and outcomes in mPFC, which appear to subserve performance monitoring in attentional set-shifting, are inherited from afferent inputs from the ACC. This hypothesis should be directly tested in future work.

The present study inspires a myriad of future questions. Under what conditions (i.e., behavioral paradigms) do efferent and/or afferent connectivity account for response heterogeneity? How will these answers vary for different subdivisions of the PFC? Finally, while the authors focused on afferent connectivity as a potential source of differences in responses across layers, there are others. Inhibitory cell types and connectivity differ by layer (Tremblay et al., 2016). Neuromodulatory axons and target receptors also vary across layers (Radnikow & Feldmeyer, 2018), and the extent to which layer-specific differences in neuromodulation account for response heterogeneity in mPFC is not well understood and is ripe for investigation.