Associations: more than connections and statistics

Not all things that resemble “A and B” are associations. Statistical learning, a phenomenon relying on repeated, unimodal inputs, is likened to associative learning by Rey, though long-lasting associative memory can develop from single trials and involve multimodal sensory inputs. Similarly, associations are here equaled to connections by Rey, but this view raises several difficulties, one of them being the risk of making any associationist account trivial, because associations would thus be present between any two (artificial or biological) connected neurons in any location of a nervous system, which clearly conflicts with our understanding of dedicated association centers, notably in mammalian or insect brains. Clarification about the concept of association is therefore needed when embarking on an associationist explanatory program, and one should be careful with oversimplification.