An empirical investigation of human dyadic systems in the time and frequency domains

This paper presents time and frequency domain analyses of psychotherapy data drawn from a unique dyadic system--the patient/therapist (P/T) system. The data are constituted by a time series defined by which member of the P/T system held the speaker role at each second of ten recorded psychotherapy consultations. The series reflects a specific property of communication within the dyad. Each series significantly followed a first order autoregressive model. That is, each was highly sensitive to the prior state of the system, or equivalently, the system showed feedforward of its prior state. These models became more significant after intervention analysis further clarified the underlying autoregressive structure. The histograms of the frequency of the length of each utterance indicated an underlying Poisson model for the emergence of speech. The parameters of the Poisson and autoregressive models were used to determine blindly which of four pairs of interviews involved the same therapist with two different patients. The method correctly identified three of the therapist four pairs but failed to identify different interviews with the same patient, a significantly negative result. These results suggest that certain time domain measures are therapist-dominated, so that the therapist drives change in these areas within the P/T system more strongly than the patient. In the frequency domain, the coherency between each of the 45 pairs of speaker series was used in an attempt to identify which time series corresponded systematically to the same therapists or patients. This proved futile because no two pairs of series--no P/T system--showed significant non-zero coherency. This result suggests that the harmonics of the switching of the speaker role are not determined by the therapist or patient alone but by the P/T system as an entity. We conclude that human dyadic systems may show predominant influence of one member or jointly determined patterns that are highly distinctive.