Does fixing nuisance variables actually resolve representational indeterminacy?

Determine whether experimental designs that hold nuisance variables constant genuinely avoid representational indeterminacy when assessing neural uncertainty representations, or whether such designs still conflate representations of uncertainty with representations of the estimated nuisance variables that can vary across trials despite fixed external conditions.

Background

To identify neural encodings of uncertainty, some studies fix external sources of uncertainty (nuisance variables) and attribute remaining variability to internal uncertainty. The authors question whether this method conclusively isolates uncertainty representations because trial-by-trial estimates of nuisance variables may still vary internally, potentially confounding results.

Their framework instead advocates testing source invariance across multiple sources of uncertainty to adjudicate whether a neural population genuinely encodes uncertainty rather than specific nuisance factors.