Model-based psychophysiology is motivated by fundamental considerations about how to draw inference on psychological – hidden – variables. It recognises that operational definitions contain loose causal models, and allows formally specifying, testing, and quantitatively applying such models. On a practical level, it can improve the temporal resolution of the analysis of slow-changing signals such as skin conductance responses (SCR), and suppresses measurement noise. The general background of model-based methods is covered in:

Bach DR & Friston KJ (2013). Model-based analysis of skin conductance responses: Towards causal models in psychophysiology. Psychophysiology, 50(1), 15-22. [PubMed] [pdf]

Models for evoked skin conductance responses (eSCR) assume that sympathetic nerve (SN) responses follow a short stimulus with constant latency. The amplitude of these responses is estimated in the framework of a general linear convolution model, using a canonical skin conductance response function (SCRF), under linear time-invariance assumptions, and an informed, linear neural model. The approach was introduced in:

Bach DR, Flandin G, Friston KJ, Dolan RJ (2009). Time-series analysis for rapid event-related skin conductance responses. Journal of Neuroscience Methods, 184, 224-234. [PubMed] [pdf]

A test of the linear time-invariance assumptions, and a development of the SCRF is published in:

Bach DR, Flandin G, Friston KJ, Dolan RJ (2010). Modelling event-related skin conductance responses. International Journal of Psychophysiology, 75, 349-356. [PubMed] [pdf]

The latest recommendations for an improved algorithm are published here:

Bach DR, Friston KJ, Dolan RJ (2013). An improved algorithm for model-based analysis of evoked skin conductance responses. Biological Psychology, 94, 490-497. [PubMed]

Models for event-related SCR (e. g. for anticipatory responses in fear conditioning) assume that the onset of the neural response is not precisely known, and estimate onset, dispersion, and amplitude of the response. They use the same SCRF as models for eSCR, but the neural model now becomes non-linear. Model inversion is therefore accomplished in the mathematical framework of Dynamic Causal Modelling (DCM), using a variational Bayes inversion scheme. This was described in:

Bach DR, Daunizeau J, Friston KJ, Dolan RJ (2010). Dynamic causal modelling of anticipatory skin conductance responses. Biological Psychology, 85, 163-70. [PubMed] [pdf]

Models for spontaneous skin conductance fluctuations (SF) are entirely uninformed about the onset of neural responses. SF are often thought to index tonic arousal. A very simple model estimates the mean number of responses times mean amplitude per time unit, as area under the curve (AUC) of the signal. The number of responses is however more informative than their amplitude, such that a more sophisticated model was developed to estimate neural response onsets and amplitudes separately. This is similar to the approach for event-related SCR: it uses a slightly modified SCRF, an uninformed non-linear neural model and the DCM model inversion framework. The AUC model, a test for LTI assumptions, and a modified SCRF for SF were introduced in:

Bach DR, Friston KJ, Dolan RJ (2010). Analytic measures for quantification of arousal from spontaneous skin conductance fluctuations. International Journal of Psychophysiology, 76, 52-55. [PubMed] [pdf]

The DCM for spontaneous fluctuations is described in:

Bach DR, Daunizeau J, Kuelzow N, Friston KJ, Dolan RJ (2011). Dynamic causal modelling of spontaneous fluctuations in skin conductance. Psychophysiology, 48, 252-257. [PubMed] [pdf]

Applications of SCRalyze:

Hayes et al. investigated the role of GABA receptors in aversive learning: Hayes DJ, Duncan NW, Wiebking C, Pietruska K, Qin P, Lang S, Gangon J, Blng PG, Verhaeghe J, Kostikov AP, Schirrmacher R, Reader AJ, Doyon J, Rainville P, & Northoff G (2013). GABA(A) Receptors Predict Aversion-Related Brain Responses: An fMRI-PET Investigation in Healthy Humans. Neuropsychopharmacology, 38, 1438-1450.

Bach & Friston have investigated negative prediction error responses in fear conditioning using DCM for event-related responses: Bach DR, & Friston KJ (2012). No evidence for a negative prediction error signal in peripheral indicators of sympathetic arousal. NeuroImage, 59, 883-884. [PubMed] [pdf]

Nicolle et al. explored autonomic responses to regret with DCM for event-related responses: Nicolle A, Fleming SM, Bach DR, Driver J, Dolan RJ (2011). A regret-induced status quo bias. Journal of Neuroscience, 31, 3320-7. [PubMed]

Bach et al. have used DCM for event-related responses as a powerful way to index fear conditioning in a small sample: Bach DR, Weiskopf N, Dolan RJ (2011). A stable sparse fear memory trace in human amygdala. Journal of Neuroscience31, 9383-9389. [PubMed] [pdf]

Talmi et al. have used a GLM approach to index sympathetic responses to pain stimuli: Talmi D, Dayan P, Kiebel SJ, Frith CD, Dolan RJ (2009). How humans integrate the prospects of pain and reward during choice. Journal of Neuroscience, 29, 14617-26. [PubMed]