Population decoding in rat barrel cortex: optimizing the linear readout of correlated population responses

Area of Science:

• Neuroscience
• Computational Neuroscience
• Sensory Processing

Background:

• Sensory information is encoded by neuronal population responses.
• Understanding how downstream neurons decode this information is crucial.

Purpose of the Study:

• To quantify the performance of a linear decoder for sensory information.
• To investigate the impact of sensory adaptation and neural correlations on decoding accuracy.

Main Methods:

• Simultaneously recorded barrel cortex neuronal responses to tactile stimuli.
• Applied signal detection theory and Fisher linear discriminant analysis to find optimal decoder weights.
• Analyzed the influence of neuronal variability, covariability, signal, and noise correlations.

Main Results:

• The optimal linear decoder significantly improved discrimination performance over simple pooling.
• Decoder performance was minimally affected by noise correlation, unlike pooling.
• Sensory adaptation enhanced decoder performance by increasing signal correlation more than noise correlation.
• An optimal decoder trained in a non-adapted state generalized well to adapted states.

Conclusions:

• A biologically plausible linear decoder can effectively decode neuronal population activity.
• Sensory adaptation impacts decoding by altering neural correlations.
• Optimal decoders can generalize across different adaptation states, offering insights into neural coding.