In Pavlovian conditioning, pairing an initially neutral stimulus (the conditioned stimulus, CS) with a biologically important stimulus (the unconditioned stimulus, US) leads the subject to anticipate the US on the basis of the CS. This is often interpreted as resulting from the creation of an internal CS-US association (associative learning). Pavlovian conditioning has proven a successful model for the development of anxiety disorders and phobias: anxiety would be the consequence of the subject anticipating an aversive US in presence of a CS.

Expression of a CS-US association can be impaired by presenting the CS on its own (extinction). This is the basis for exposure therapy of anxiety and phobia. Extinction is unfortunately susceptible to recovery effects. Notably, if the original CS-US pairing occurs in a context A while extinction occurs in a context B, the CS will trigger anticipation of the US (and hence anxiety) if presented in the original acquisition context A or even in a new context C. This phenomenon, known as renewal, is an important model of relapse after exposure therapy or drug rehabilitation treatment. For this reason, there has been a considerable research effort in recent years to discover how to mitigate renewal.

My current work falls in this line of research. With my collaborators (quite often, Dr. Ralph Miller from SUNY-Binghamton), we aimed at comparing the efficiency and the susceptibility to renewal of various forms of nonreinforcement (the CS is presented by itself. Extinction falls in this category) and associative interference (the CS is paired with another stimulus). We mainly use the streaming procedure, a relatively recent procedure for the study of human associative learning developed by Allan and collaborators at McMaster University, in which participants are exposed to very rapid streams of stimuli, each stream corresponding to a condition. Because streams are so short, the streaming procedure allows for fully within-subject. It also makes it ideally suited for online studies, which allows to use it with large samples. The combination of these two features means that the streaming procedure is able to reach considerable statistical power (Maia et al., 2018; Jozefowiez, 2020; Jozefowiez & Miller, 2022; Jozefowiez et al., 2021). Our research has shown so far that the comparative efficiency/susceptibility to renewal of nonreinforcement and associative interference depends on which aspect of learning is assessed (expectancy learning, conscious expectations resulting from conditioning, vs. evaluative conditioning, change in the valence of the CS following conditioning. See Jozefowiez et al, 2020, in press), the CS-US contingency (Jozefowiez et al., 2023), as well as complex interplay between the valence of the US originally paired with the CS and the one used to induce associative interference (Jozefowiez et al., forthcoming).

Previously, I worked on temporal cognition in men and animals, looking at the interaction between reward processing (reinforcement) and time perception (Jozefowiez et al., 2005, 2006, 2013, 2018) which led to the development of a model (BEM for Behavioral Economic Model. Jozefowiez et al., 2009a) with a surprising relevance for studies aimed at demonstrating metacognition (cognition about cognition) in animals (Jozefowiez et al, 2009b).