Implementation of structure-mapping inference by event-file binding and action planning: a model of tool-improvisation analogies

Abstract

Structure-mapping inferences are generally regarded as dependent upon relational concepts that are understood and expressible in language by subjects capable of analogical reasoning. However, tool-improvisation inferences are executed by members of a variety of non-human primate and other species. Tool improvisation requires correctly inferring the motion and force-transfer affordances of an object; hence tool improvisation requires structure mapping driven by relational properties. Observational and experimental evidence can be interpreted to indicate that structure-mapping analogies in tool improvisation are implemented by multi-step manipulation of event files by binding and action-planning mechanisms that act in a language-independent manner. A functional model of language-independent event-file manipulations that implement structure mapping in the tool-improvisation domain is developed. This model provides a mechanism by which motion and force representations commonly employed in tool-improvisation structure mappings may be sufficiently reinforced to be available to inwardly directed attention and hence conceptualization. Predictions and potential experimental tests of this model are outlined.

From “Oh, OK” to “Ah, yes” to “Aha!”: Hyper-systemizing and the rewards of insight

“…Hyper-systemizers would be expected to perform better than "balanced" or mentalizer controls of matched verbal ability on semantic-association tests as employed in insight studies (Jung-Beeman et al, 2004;Kounios et al, 2007;Sandkuhler and Bhattacharya, 2008) and on standard verbal analogies (Gentner, 2003;Holyoak, 2005). Hyper-systemizers would also be expected to perform better than balanced or mentalizer controls of matched visuo-spatial ability on toolimprovisation problems, which are effectively visuo-spatial analogies (Fields, 2010). Conversely, hyper-mentalizers would be expected to perform no better than balanced or systemizer controls on mentalizing tasks, such as correctly inferring the intentions of actors in stories.…”

Motion as manipulation: implementation of force–motion analogies by event-file binding and action planning

“…Both this analogy and the others that appear explicitly in Rutherford (1911) are surprisingly similar to tool-improvisation analogies. It is shown here that these latter analogies can be implemented using the representational and inferential apparatus provided by the ''eventfile manipulation'' (EFM) model developed in Fields (2011a). Based on these EFM model implementations, it is suggested that electrons:nucleus::planets:sun is derivative from alpha-particle:central-charge::electron:atom, and that it is compelling not because it rests on the manipulation of concepts such as REVOLVE-AROUND, but because it rests on an ancient and ubiquitous motor memory of orbital motion.…”

“…In the almost three decades since the introduction of structure mapping as a formal model of analogy by Gentner (1983), a ''standard view'' of structure-mapping inference as mechanistically dependent on the recognition of explicit lexical symbols representing relational concepts, and hence as a humanspecific inferential capability, has come to dominate the analogy research community. This standard view has, however, recently been challenged by the observation that tool-improvisation analogies, which are carried out in the wild by many non-human animals, involve structuremapping inferences over force-motion relations (Fields 2011a).…”

“…Mammals and birds of many species improvise tools; tool-improvisation analogies must, therefore, be implemented by neurocognitive mechanisms that do not depend on human-like natural language and, if the preponderance of evidence is to be accepted, do not depend on explicit ''mentalese'' encodings or conscious awareness of relational concepts (Penn and Povinelli 2007;Suddendorf and Coraballis 2007;Penn et al 2008). On the basis of primarily neurofunctional evidence, Fields (2011a) proposed that tool-improvisation analogies are implemented across phyla by event-file binding and action planning systems that are structurally homologous across mammals and appear to be shared as functional systems by mammals and birds. The EFM model formalizes this proposal, showing how the complex relation ''Tool A moved along trajectory B will transfer force C to target D, thus achieving goal E'' can be implemented by a pre-motor action planning network capable of retrieving, comparing, and recombining components of memory-resident representations of multiple instances of successful and unsuccessful actions.…”

“…For example, the inferences supporting tool improvisation require for practical validity the mapping of relationships between applied forces and achieved motions from remembered situations to novel situations, typically independently of the surface features of the objects involved; such inferences therefore meet the definitional criteria for structure-mapping analogies (Fields 2011a). Mammals and birds of many species improvise tools; tool-improvisation analogies must, therefore, be implemented by neurocognitive mechanisms that do not depend on human-like natural language and, if the preponderance of evidence is to be accepted, do not depend on explicit ''mentalese'' encodings or conscious awareness of relational concepts (Penn and Povinelli 2007;Suddendorf and Coraballis 2007;Penn et al 2008).…”