Marion Clara Vorms

Models are generally used by scientists to obtain predictions and to provide explanations about phenomena. Their predictive and explanatory power is generally thought of as depending on their representative power. It is still not clear, though, in virtue of which features models allow scientists to draw inferences about the system they stand for. In this paper, I focus on a special kind of models, namely imaginary models (I-models) such as the simple pendulum. The main question I address is: how do scientists use I-models in representing target systems? First, I propose a clarification of the very notion of representation, by emphasizing the importance of what I call the format of a representation to the inferences cognitive agents can draw from it. Then, I analyze the various representational relationships that are in play in the use of I-models. I finally conclude that there is no special semantics to be applied to I-models, and that the study of the representational power of models in general should instead focus on the variety of the formats that are used in scientific practice.

Gergely and Csibra's theory, known as "natural pedagogy theory", is meant to explain how infants fast-learn generic knowledge from adults. In this paper, my goal is to assess the explanatory import of this theory in a particular case, namely the phenomena known as "A-not-B errors". I first propose a clarification of natural pedagogy theory's fundamental hypotheses. Then, I describe Topál et al.'s (Science, 321, 1831-1834, 2008) experiments, which consist in applying natural pedagogy theory's framework to the A-not-B errors. Finally, I show that natural pedagogy theory, in its actual stage of development, does not suffice to choose between various interpretations of Topál et al.'s experimental results

It is quite widely acknowledged, in the field of cognitive science, that the format in which a set of data is displayed (lists, graphs, arrays, etc.) matters to the agents' performances in achieving various cognitive tasks, such as problem-solving or decision-making. This paper intends to show that formats also matter in the case of theoretical representations, namely general representations expressing hypotheses, and not only in the case of data displays. Indeed, scientists have limited cognitive abilities, and representations in different formats have different inferential affordances for them. Moreover, this paper shows that, once agents and their limited cognitive abilities get into the picture, one has to take into account both the way content is formatted and the cognitive abilities and epistemic peculiarities of agents. This paves the way to a dynamic and pragmatic picture of theorizing, as a cognitive activity consisting in creating new inferential pathways between representations.