Qualitative Process theory provides a formal representation for human-like models of continuous processes. Prior research mapped qualitative process elements onto English language constructions, but did not connect the representations to existing frame semantic resources. Here we identify and classify QP language constituents through their instantiation in FrameNet frames to provide a unified semantics for linguistic and non-linguistic representations of processes. We demonstrate that all core QP relations can map to FN, though larger QP evoking phrasal constructions do exist outside of this mapping. We conclude with a corpus analysis showing that these frames occur in natural text involving a variety of continuous processes.

Apperly and Butterfill (2009) argue that adult theory of mind (ToM) requires two parallel systems. One system, efficient but inflexible, enables rapid judgements by operating without explicit modeling of beliefs, while a separate, effortful system, enables richer predictions over more complex belief encodings. Here, we agree with their qualitative distinction but propose a different model: a single process, but with effortful re-representation leading to two phases of ToM reasoning. Efficient reasoning, in our view, occurs over representations that include actions, but not necessarily explicit belief states. Effortful reasoning, then, involves re-representation of these initial encodings in order to handle errors, resolve real-world conflicts, and fully account for others’ belief states. We present an implemented computational model, based in memory retrieval and structural alignment, that illustrates our approach.

Many psychologists have argued that language acquisition plays an important role in the development of Theory of Mind (ToM) reasoning in children. Several accounts of this interaction exist: some believe that language gives children the ability to express already formed ToM reasoning (e.g. He, Bolz, & Baillargeon, 2011), while others argue that learning specific grammatical structures engenders new reasoning abilities (e.g. de Villiers & Pyers, 1997). Questions remain about the mechanism by which this interaction occurs. In this paper, we show that the Analogical Theory of Mind (AToM; Rabkina et al., 2017) computational model can bootstrap aspects of ToM reasoning from sentential complement training, and that its performance matches improvement patterns of children who are trained using similar stimuli. This provides an implemented algorithmic account of bootstrapping ToM reasoning from language within a broader model of ToM development.

The creativity of natural language poses a significant theoretical problem. One example of this is denominal verbs (those derived from nouns) such as spoon in “She spooned me some sugar”. Traditional generative approaches typically posit a unique entry in the lexicon for this usage, though this approach has difficulty scaling. Construction Grammar has evolved as a competing theory which instead allows the syntactic form of the sentence itself to contribute semantic meaning. However, how people learn syntactic constructions remains an open question. One suggestion has been that they are learned through analogical generalization. We evaluate this hypothesis using a computational model of analogical generalization to simulate Kaschak and Glenberg’s (2000) study regarding interpretation of denominal verbs.

Several theories about Theory of Mind (ToM) have beenproposed. The most well-known of these are Theory Theoryand Simulation Theory, although alternative and hybridtheories do exist. One such theory, proposed by Bach (2011,2014), is based on the Structure-Mapping theory of analogy,which has been shown to play a key role in cognitivedevelopment. There is evidence that children are more likely topass false belief tasks when trained using stories that are easyto compare via structural alignment, as opposed to stories thatare difficult to compare in this way (Hoyos, Horton & Gentner,2015). This paper shows how a computational model based onBach’s account can provide an explanation for the Hoyos et al.training study and proposes directions for future research onhuman subjects.