On the origins of abstract combinatorial thought: composing meaning in infancy

Talk from Pr. Agnes Melinda Kovacs Human adults can productively combine existing concepts and thoughts to generate an unbounded sequence of novel thoughts, as also reflected by natural language production. One interesting debate, however, concerns the question whether pre-verbal or nonverbal creatures, who do not have access to a fully-fledged language system can rely on representations and mental operations that allow forming abstract concepts and flexibly combining them using logical operators. In this talk I will present three lines of studies with infants in which we try to address some of these questions. First, we aimed to tackle on early combinatorial abilities in language and beyond language. In one set of studies, we taught novel quantity labels (for ‘one’ and ‘two’) to 12-month-olds and found that they could flexibly combine these newly learned labels with familiar nouns. In a further study we targeted non-verbal combinatorics and observed that infants could combine functions by learning two arbitrary operations and combining their outcomes. In a second series of experiments, we found evidence that already at 18 months the early meanings of verbal negation are truth-functional, and infants map different forms of negation to propositional negation. They readily interpreted the negation particle in composition with other elements, which then served as input for logical inferences. In a third line of study, we targeted structured representations in the visual domain and found that infants, similarly to adults, can extract the gist of complex natural scenes, and that conceptual information seems to structure their scene representations. These studies suggest that infants are particularly prone to rely on abstract concepts and flexibly combine information from different levels to form structured representations and generate new meaning.​