Pierre Barrouillet

Research interests

My research interests were initially oriented towards the development of logic and categorization within a piagetian perspective (Barrouillet & Poirier, 1997). My current research is no longer devoted to these topics and focuses on three main domains: conditional reasoning and its development, working memory, and cognitive arithmetic.


Conditional reasoning is reasoning permitted by sentences involving the logical connective If (e.g., If I worked on psychology of reasoning, then I would become crazy or I’m ready to work with you only if you don’t talk to me about reasoning). About ten years ago, I initiated with Jean-François Lecas a research on the way children and adolescents represent conditional sentences and how they reason from them. We mainly used tasks in which children had to identify those cases that are incompatible with a given conditional sentence (e.g., if the piece is square, then it is red) or to construct those cases that are compatible with these sentences. Many of our studies revealed a clear developmental trend fairly compatible with what Phil Johnson-Laird proposed in his Mental Models theory of conditional in adults (Barrouillet, Grosset & Lecas, 2000; Barrouillet & Lecas, 1998, 1999, 2002; Barrouillet, Markovits, & Queen, 2001). Young children usually consider only one case as compatible with the conditional sentence (i.e., squares that are red in the previous example). This primitive interpretation is progressively enriched with an additional representation of the case in which the two terms are falsified (e.g., circles that are blue) and finally by a third representation of pieces that are not square but that are red. Thus, we demonstrated that the development tends towards a representation conforming to the three-model representation postulated by Johnson-Laird’s mental models theory in adults. However, developmental studies revealed also phenomena that do not fit very well with the standard mental models theory. Henry Markovits and I proposed a modified mental models theory of the development of conditional that accounts for several content and context effects in children (Markovits & Barrouillet, 2002, 2004) and adults (Grosset, Barrouillet, & Markovits, 2005). More recently, I proposed a further modification of the mental models theory that reconciles the mental models approach with the suppositional theory of conditional put forward by Jonathan Evans and David Over (Barrouillet, Gauffroy, & Lecas, submitted). This recent modification opens an avenue for future research on development, Caroline Gauffroy being in charge of this program in our lab.


Working memory is a capacity-limited system devoted to the simultaneous maintenance and treatment of information usually considered as being mainly implicated in high level cognition. Thus, I first studied the implication of working memory capacities in conditional (Barrouillet & Lecas, 1999) and transitive reasoning (Barrouillet, 1996; Favrel & Barrouillet, 2000). Some years ago, I initiated a collaboration with Valérie Camos from the Université de Bourgogne to study working memory development (Barrouillet & Camos, 2001). The striking results we obtained led us to propose a new model of working memory named the Time-Based Resource-Sharing model (TBRS, Barrouillet, Bernardin, & Camos, 2004; Barrouillet & Camos, 2007). According to this model, the complex activities used as processing components within the traditional WM span tasks, such as reading comprehension, equation solving or counting, can be analyzed as a series of elementary steps that capture attention for short periods of time between which attention can be diverted and intermittently switched away for maintenance purpose. Indeed, information suffers from a time-related decay when attention is switched away, and frequent refreshing by attentional focusing is needed.

This theory gave us the rationale for developing a series of WM span tasks in which the processing components consist of a succession of very simple activities performed under time constraints such as reading digits, adding or subtracting 1 or 2 to single-digit numbers (the reading digit span and the continuous operation span tasks, Barrouillet et al., 2004). In line with the TBRS model predictions, when performed under time constraints, these elementary activities proved to be as disruptive as complex tasks for concurrent maintenance of information (Lépine, Bernardin, & Barrouillet, 2005), and our new WM span tasks proved to be better predictors of academic achievement than the traditional reading span and operation span tasks (Lépine, Barrouillet, & Camos, 2005). These results confirmed Barrouillet et al. (2004) claim that cognitive cost is not a matter of complexity per se. Very simple activities consume cognitive resources as far as they require attention. Thus, the TBRS model is also a theory of cognitive load conceived of as the proportion of time during which a given activity captures attention and impedes concurrent processing such as memory refreshing (Barrouillet et al., 2007).

We are now developping many research programs inspired by the TBRS model. Under the direction of Valérie Camos, Sophie Portrat at the Université de Bourgogne is exploring the relationships between the capture of attention by executive functions and cognitive cost. In our lab, Evie Vergauwe is in charge of the extension of the TBRS model to visuo-spatial cognition, whereas Vinciane Gaillard is in charge of an FNRS program about the development of working memory within the TBRS model in collaboration with Valérie Camos, Chris Jarrold at the University of Bristol and Nelson Cowan at the University of Missouri. We are also working on the implications of the model concerning short-term memory and serial recall. Valérie Camos is developing a research program about the relationships between central (attention-based) and peripheral (representation-based) interferences within working memory as well as a computer simulation of the model. The TBRS model is the first step towards a broader theory of cognition we plan to develop in the future.


Cognitive arithmetic and numerical cognition constitute my third theme of research. I collaborated with Valérie Camos during her PhD thesis in which she demonstrated that, very early in development, counting is an integrated activity that does no longer require a demanding coordination between pointing and saying (Camos, Barrouillet, & Fayol, 2001; Camos, Fayol, & Barrouillet, 1999). However, she demonstrated that counting involves a cognitive cost even in adults (Camos & Barrouillet, 2004). This interest in numerical activities led us to develop a new model of number transcoding we named ADAPT (A Developmental Asemantic and Procedural Transcoding model) that has been published in 2004 in Psychological Review. The main strength of this model is that it accounts for learning and development as well as for the difficulties encountered in number transcoding by adolescents with learning disabilities and brain-damaged patients. I also studied word problem solving with Catherine Thevenot who proposed a mental model account of this activity (Thevenot, Barrouillet, & Fayol, 2004 ; Thevenot, Devidal, Barrouillet, & Fayol, 2007).

Most of my work on numerical activities concerns cognitive arithmetic and the move from algorithmic strategies to direct retrieval in addition and subtraction problem solving (Barrouillet & Fayol, 1998 ; Fayol, Barrouillet, & Roussel, 2002). I recently examined the relationships between this developmental change in addition and the WM capacities in children (Barrouillet & Lépine, 2005) and I’m currently addressing the same question in subtraction. Current projects concern number transcoding in collaboration with Valérie Camos and number encoding with Catherine Thevenot (Thevenot & Barrouillet, 2006).


Other topics have concerned children’s drawing as well as spelling in collaboration with Michel Fayol who were formerly at the Université de Bourgogne and who is now at the Université Blaise Pascal in Clermont-Ferrand.