The empirical investigation of unobservable psychological processes usually rests on operational definitions. As an alternative, we propose the use of explicit causal models. This is particularly useful in psychophysiology, where formal models can be expressed mathematically, exploiting biophysical constraints, and inverted to yield estimates of unobservable processes. In psychophysiology, recent advances have been made in causal modeling for skin conductance responses, which we discuss to exemplify the development of such models. Empirical evidence suggests that these methods have a greater validity compared to operational approaches. This review concludes by considering the theoretical implications for the field of psychophysiology and benefits for practical data analysis.