Parsing peak provoked craving


Sayette & Tiffany [1] have produced a thoughtful response to the crisis in cue–reactivity (CR) research that was triggered in part by Perkins' [2] challenge to CR's clinical relevance. The suggestion that we study craving states generally, with less concern about parsing what part of the craving is attributable to cues and what part to other factors, is well taken and has already been productive, as evidenced by the cited Sayette work documenting the effects of craving on emotion and cognition.

The proposed position deserves careful parsing. It seems to include three ideas, corresponding to the three words in the peak provoked craving (PPC) approach. They are discussed below, taking the terms in reverse order.


The PPC model emphasizes that the overall level of craving matters, regardless of source. This has the dual benefit of being parsimonious and also consistent with the data showing that craving intensity is related to outcomes such as smoking and dependence, regardless of source [3, 4]. While the proposition that craving intensity matters may seem trivial, we should remember that some authors [5] have suggested that craving is epiphenomenal and perhaps unimportant, so the proposition is not trivial.


It seems surprising that ‘provocation’ is so prominent in the proposal because, in the PPC paradigm, provocation (i.e. exposure to cues) does not seem to matter greatly—it is simply a means of raising craving to interesting levels, and may not even be necessary if craving is already intense. The PPC approach disavows a key goal of the CR approach—the desire to model the impact of cues (more on this below).


The model suggests that some features or effects of craving appear only when craving is very intense—at its peak; i.e. that very intense craving differs in some qualitative way (not just quantitatively) from cravings of lesser intensity. Yet evidence for this is scant. In fact, several of the papers cited show that craving has a linear relationship with important variables such as dependence [3] (although see [6]); i.e. that its effects are roughly constant across the spectrum of intensity. Indeed, where non-linearity has been observed, it often is the opposite of what is proposed. For example, we have observed, both in real life [7] and in the laboratory [8], that as craving reaches peak values, its effects on smoking decrease substantially and flatten out, such that observing craving only near its peak values actually blunts and masks the relationship between craving and smoking. The idea that ‘peak’ craving states differ in some qualitative way from less-intense craving states is an interesting (provocative?) one, and therefore one that demands clear empirical evidence—evidence that is not yet there. One of the most valuable contributions of the PPC model, then, may be to stimulate (cue?) researchers to address this question.

In thinking how the PPC approach differs from the current CR approach, it seems important to recall what motivated CR research. The interest in the effects of cues was stimulated by the observation that cues often seem to trigger lapses (e.g. [8]), and the hypothesis that learning processes had imbued certain stimuli with the power to trigger drug use. Thus, understanding the specific effects of cues—not craving, per se—was the central focus. Indeed, the nearly exclusive focus on craving as an outcome of cue exposure, and its corresponding neglect of drug use as an outcome, has rightly been criticized [2]. A second focus of CR research has been on individual differences, on the presumption that some individuals are meaningfully more reactive to cues in general or to certain cues in particular (as might be expected from varying learning histories), with implications for vulnerability to relapse. Thus, CR studies treat the levels of craving resulting from a cue exposure as important indicators of individual differences, where the PPC approach, by focusing on only those cases where peak levels are achieved, would seem to compress or dismiss individual variations. Importantly, Sayette & Tiffany do not suggest that the PPC approach supplant the CR approach, but rather supplement it.

Sayette & Tiffany deserve our thanks for prompting renewed examination of the dynamics of craving and the methods we use to study it.

Declarations of interest

The author consults to GlaxoSmithKline Consumer Healthcare exclusively on matters related to smoking cessation, and has an interest in a company developing nicotine medications.