TL;DR
Researchers have uncovered a specific "opponent inhibition" wiring pattern in the posterior parietal cortex (PPC) that is crucial for decision-making. This motif, where excitatory neurons preferentially connect to similarly selective inhibitory neurons, and inhibitory neurons connect to oppositely selective excitatory neurons, helps sharpen neural representations of choices and improve decision accuracy.
What Problem Does This Paper Address?
The posterior parietal cortex (PPC) is known to be involved in decision-making, showing "choice-selective activity" – meaning different neurons become active depending on the choice a subject is about to make. However, the precise neural circuitry and synaptic connections that give rise to this selective activity have remained largely unknown. Specifically, how the physical wiring of neurons in the PPC supports this functional selectivity during complex tasks was a major open question.
Methods
The study employed a multi-modal approach:
- Virtual Reality Behavior & Two-Photon Calcium Imaging: Mice were trained on a two-alternative forced-choice task in a virtual reality T-maze. Their neural activity in the PPC was recorded using two-photon calcium imaging to identify neurons with choice-selective activity.
- High-Throughput Electron Microscopy (EM): Immediately after imaging, the brains were preserved, and the same neurons were then imaged using advanced serial-section EM. This allowed for the reconstruction of the 3D structure of individual neurons and, crucially, the identification of their synaptic connections at nanometer resolution.
- Circuit Modeling: Both simplified linear rate models and more complex recurrent neural network (RNN) models were used to simulate the effects of the identified connectivity motifs on neural activity and decision-making performance.
This combination is powerful because it directly links functional activity (calcium imaging) to structural connectivity (EM) within the same individual neurons.
Datasets
The study generated a novel dataset by combining in vivo functional imaging data with high-resolution EM structural data from the same set of neurons in mouse PPC. This dataset allowed for the direct correlation of neuronal activity and synaptic connectivity. For comparison, they also analyzed a previously generated EM volume of the visual cortex (V1).
Key Insights and Ideas
- Opponent Inhibition Motif: The core finding is the identification of a specific "opponent inhibition" motif in the PPC. This isn't just random wiring; it's a structured pattern:
- Excitatory-to-Inhibitory (E-to-I): Excitatory neurons tend to connect to inhibitory neurons that share the same choice selectivity.
- Inhibitory-to-Excitatory (I-to-E): Inhibitory neurons tend to connect to excitatory neurons that have the opposite choice selectivity.
- PPC vs. V1 Connectivity: This selective E-to-I connectivity is significantly different from what's observed in the visual cortex (V1), where E-to-I connections are generally less selective. This suggests that association areas like PPC might have distinct wiring principles compared to primary sensory areas.
- Functional Significance of Synapse Size: The size of excitatory synapses (measured by PSD area) correlated with selectivity. Co-selective E-to-I synapses were larger than anti-selective ones, suggesting that synapse strength is tuned to selectivity.
- I-to-E Selectivity Driven by Dendrite Targeting: The anti-selectivity of I-to-E connections was primarily driven by synapses targeting basal dendrites of pyramidal neurons, suggesting that different dendritic compartments might play specialized roles in processing choice information.
- Model Validation: Both linear and RNN models confirmed that this opponent inhibition motif amplifies choice-selective signals, enhances competition between opposing neural populations, and ultimately improves the accuracy of decision-making, especially in the presence of noise.
Key Results
- E-to-I Connectivity: Local E-to-I connectivity rates were significantly higher than E-to-E in PPC. Co-selective E-to-I connections were more frequent and had larger synapse sizes (PSD areas) compared to anti-selective ones.
- I-to-E Connectivity: Local I-to-E connectivity rates were more than three times higher than E-to-I. Normalized synapse frequency was slightly higher for anti-selective I-to-E pairs, indicating inhibitory neurons preferentially target oppositely selective excitatory partners.
- Model Performance: Circuit models incorporating the opponent inhibition motif showed enhanced separation of neural activity representing different choices and improved decoding accuracy of trial type, particularly when opponent inhibition was stronger.
Limitations and Future Work
- Sample Size: The study acknowledges that the EM volume, while large for connectomics, still represents a "modest sample size" of functionally characterized neurons and synapses. Larger datasets could reveal additional circuit motifs.
- E-to-E Connectivity: The dataset was too sparse to confidently evaluate E-to-E selectivity, which is a common motif in many decision-making models. Future work with larger volumes might address this.
- I-to-I Connectivity: The number of functionally characterized inhibitory neurons was insufficient to assess I-to-I connectivity.
- Plasticity: The study hints at plasticity, showing that structure-function correlations weaken with time. Future work could explore how these connectivity motifs change over time or during learning.
- Generalizability: While the study focuses on PPC, future research could investigate if similar opponent inhibition motifs are present in other decision-making circuits or brain areas.
Summary updated