The study tested the hypothesis, derived from Teasdale and Barnard's (1993) Interacting Cognitive Subsystems (ICS) model, that two commonly used techniques in cognitive therapy, automatic thought records (ATRs) and behavioural experiments (BEs), would have their primary impact on different cognitive subsystems. Teasdale (1997a, b) has suggested that while ATRs tend to impact on a logic-based propositional information processing system, BEs impact on a more holistic implicational system, which has extensive links with emotion. Quantitative and qualitative data derived from three groups of participants undertaking cognitive therapy training courses, who practised ATRs and BEs on themselves, supported Teasdale's theory. Differences were found in participants' ratings of behaviour and belief change, with BEs being perceived as more powerful and compelling than ATRs, as predicted by the theory. There were also qualitative differences in participants' experiences of the two techniques, suggestive of different modes of processing. The data indicated the particular importance of the therapist's role in BEs. In order to provide a more comprehensive account of ATRs and BEs, it was suggested that Teasdale's theory could profitably be extended to include the role of the therapist in promoting adaptive oscillation between the implicational and propositional modes.