**Definition:** Pre-coloring replaces unknown fNIRS signal autocorrelation with autocorrelation of a known form. This involves the use of temporal smoothing with a smoothing kernel of known correlation structure. The reason for precoloring is to enforce that the signal complies with the assumption of the model that is used for statistical inference. For instance, the ordinary least squares solution to the general linear model assumes that the residuals are serially independent. If the residuals are serially dependent – as in the case of the fNIRS signal due to e.g. physiological fluctuations-, however, this will result in the underestimation of estimates of parameter standard deviations and thus overestimation of statistical significance. Please note that, precoloring, if not used properly, can lead to an overestimation of the true effect. Definition: Pre-coloring replaces unknown fNIRS signal autocorrelation with autocorrelation of a known form. This involves the use of temporal smoothing with a smoothing kernel of known correlation structure. The reason for precoloring is to enforce that the signal complies with the assumption of the model that is used for statistical inference. For instance, the ordinary least squares solution to the general linear model assumes that the residuals are serially independent. If the residuals are serially dependent – as in the case of the fNIRS signal due to e.g. physiological fluctuations-, however, this will result in the underestimation of estimates of parameter standard deviations and thus overestimation of statistical significance. Please note that, precoloring, if not used properly, can lead to an overestimation of the true effect.

**Alternative definition:**

**Synonym: **

**References:** https://doi.org/10.1002/1097-0193(200102)12:2%3C61::AID-HBM1004%3E3.0.CO;2-W

**Related terms:** Pre-whitening, noise, GLM