Modelling event-related skin conductance responses

Abstract

Analytic tools for psychophysiological signals often make implicit assumptions that are unspecified. In developing a mathematical framework for analysis of skin conductance responses [SCRs], we formalise our assumptions by positing that SCRs can be regarded as the output of a linear time-invariant filter. Here, we provide an empirical test of these assumptions. Our findings indicate that a large component of the variance in SCRs can be explained by one response function per individual. We note that baseline variance (i.e. variance in the absence of evoked responses) is higher than variance that could not be explained by a linear time-invariant model of evoked responses. Furthermore, there was no evidence for nonlinear interactions among evoked responses that depended on their temporal overlap. We develop a canonical response function and show that it can be used for signals from different recording sites. We discuss the implications of these observations for model-based analysis of SCRs.

Fig. 1

Variance partitioning and response functions were estimated for different stimulus classes. Top panel: bar charts show variance components across subjects; that is, variance explained by one common response function (black), between-subjects variance (grey) and residual variance (white) for each experiment. Light grey: baseline variance in the absence of evoked responses for the visual detection experiment. Box plots depict explained variance within subjects, showing median (line), quartiles (box), and range (whiskers). Outliers (values outside the 1.5 × interquartile range) are shown as individual dots; whiskers then represent the data within the interquartile range. Bottom panel: empirical (PCA) response functions across participants for the different experiments.

Fig. 2

A canonical response function was derived from observed responses (shown as the first PCA component ± standard deviation across all observations), and was described analytically as an exponentially modified Gaussian function (top panel). Time and dispersion derivatives were constructed to account for differences in response shape between participants and experimental conditions (middle panel); the complete basis set could explain 64.0% of the total variance. Latency and rise time are comparable to the function proposed by Lim et al. (1997).

Fig. 3

The linearity assumption was tested by presenting either single white noise sounds, or a sequence of two sounds with differing ISIs, separated by silent periods of 30–40 s. For each individual, a response function was estimated from their responses to single events, and responses to double events were estimated under time-invariance assumptions. It turns out that responses to second events are smaller than to the first event in each sequence, regardless of ISI (2–9 s); however, there is no evidence for dependency of the repetition suppression on ISI. Averaged residuals are similar for the first and second event, thus indicating no systematic alterations of response shape at different ISIs.

Fig. 4

Variance portioning and response functions for hand (thenar/hypothenar), finger (volar surface, 2nd/3rd finger, middle phalanx), and medial foot (medial plantar) recordings. The ratio of explained variance is similar, and the response functions mainly differ in peak latency. Note that the canonical response function depicted in Fig. 2 can explain responses from all three recording sites almost equally well.