Seminar – MPFI Research Seminar

Be a Splitter, not a Lumper! On the Proper Granularity to Study the Neurobiology of Cognition and Language

David Poeppel

Feb 12,
  • Time:
  • Location: MPFI - Elmore Auditorium
  • Host: David Fitzpatrick
  • Speaker: David Poeppel

In the spirit of convenient simplification, most of us can be classified as splitters or lumpers with respect to our research approaches. If animal systems are to play any role in understanding human language and cognition, I argue that being a lumper will doom the work from the get-go. Working with categories that are too broad (e.g. “semantics”) will render the comparative results — at best — correlational. If we embrace, instead, ‘radical decomposition’ as a research strategy, i.e. diligently focusing on identifying the ‘elementary particles’ (or the computational primitives) of the systems as well as their interactions, there is hope that we may understand the system at the right level of granularity and generate explanatory understanding. Experimental examples from the neurobiology of language will illustrate this conjecture. There exists, in any case, reason to remain naively optimistic: research on animal communication systems can help us part of the way already, because for certain questions (i.e. the sensorimotor interfaces) we can use approaches with appropriate spatial resolution and temporal resolution to understand various subroutines. But unless we have the appropriate ‘conceptual resolution,’ the relation between animal and human cognitive systems is likely to remain more metaphoric than mechanistic – and some of the deeper questions will remain opaque.