Pratik Sachdeva

Resolving Non-identifiability Mitigates Bias in Models of Neural Tuning and Functional Coupling

The paper identifies and mitigates bias in neural data models by resolving non-identifiability, showing that coupling is overestimated and tuning is underestimated in common methods.

Not optimal, just noisy: the geometry of correlated variability leads to highly suboptimal sensory coding

The paper shows that correlated variability in neural populations leads to highly suboptimal sensory coding according to novel null models.

Heterogeneous synaptic weighting improves neural coding in the presence of common noise

The paper shows that heterogeneous synaptic weights can improve neural coding in the presence of common noise by enhancing Fisher and Shannon information, even when amplifying noise.

Accurate inference in parametric models reshapes neuroscientific interpretation and improves data-driven discovery

The paper applies an improved statistical inference framework to neuroscience models, finding more accurate and sparse parameter estimates that reshape interpretations across multiple datasets.