Skip to main content
European Commission logo print header

The Syntax of the Mind: A Comparative Computational Approach

Final Report Summary - SOMACCA (The Syntax of the Mind: A Comparative Computational Approach)

Compared to animal communication systems, human language is unparalleled in its complexity, flexibility and specificity. Furthermore, language appears to require certain cognitive capabilities that are unique in the animal world. Nonetheless, many other components of language are shared with other animals, ranging from simple aspects of auditory perception or vocal production, to quite complex cognitive abilities of pattern perception, learning and memory. A primary goal of the SOMACCA grant was to determine which component mechanisms involved in language are shared with other animals and, by process of elimination, which are not.

We explored this by performing a rigorous comparison of humans and animals on a broad suite of cognitive tasks. This required the development of new and innovative ways to probe cognition in animals, and also in humans without using language. SOMACCA focused mainly on complex pattern perception – music and language - areas where humans seem to differ most from other animals, but we also explored more basic auditory perception and vocal production capabilities. Formal language theory, a branch of the mathematical theory of computation, was used to provide a rigorous theoretical framework. A key empirical approach was artificial grammar learning: A set of stimuli are generated following some abstract rule(s), and after exposure to these stimuli, human or animal observers are tested with novel stimuli, to determine what rules they have extracted. The rules acquired may or may not be the same as those used to generate the stimuli, and this allows us to explore the proclivities and constraints on rule learning in different species. The SOMACCA team explored learning of both visual and auditory patterns in a wide variety of species including humans, chimpanzees, monkeys, parrots, pigeons, and ravens. A surprise finding is that participating in cognitive tasks reduces stress in squirrel monkeys, as revealed by a drop in cortisol after performing our study. This has positive implications for cognitive enrichment for zoo or laboratory animals, or pets. Comparisons between species constitute the first sense in which the project is “comparative”

Another type of comparison concerned cognition within the human species, across the different domains of language, music and visual patterns. Although these domains are often treated as fundamentally different, the SOMACCA grant explored the possibility that there are fundamental similarities among them. One fundamental similarity involves a cross-domain mechanism for learning and applying abstract, hierarchical rules, which can apply to music, language and art in similar ways. To test this hypothesis, we compared human learning of auditory and visual patterns, including spoken and musical stimuli as well as simple animated “movies” to isolate similarities and differences across sensory domains. We have successfully demonstrated that humans possess an abstract capacity to both recognize and generate grammatical and recursive structures in the visual domain, and compared recursion and other forms of hierarchy (e.g. phonological stress) in music and speech.

The results from this five-year project support the hypothesis that humans have an unusual cross-modal ability to process abstract patterns in hierarchical terms (involving tree structures) regardless of domain or sensory modality. This “Dendrophilia Hypothesis” suggests that humans have a general capacity and proclivity for hierarchical cognition – a love of trees - spanning different sensory and cultural domains. We explored the aesthetic implications of this hypothesis by allowing non-artists to generate complex abstract patterns in the lab using creative “FlexTiles” software. Results suggested a general proclivity for particular types of order and complexity in our species. Further, our animal research suggests that even simple relational or hierarchical tasks are very challenging for animals, even large-brained intelligent species such as the kea, a New Zealand parrot species. These data support the supposition that human hierarchical abilities are at least unusual, if not unique, in the animal world. In contrast, we found that simpler sequential patterns (e.g. patterns typical of phonology) were graspable by multiple species of bird and nonhuman primates. Future comparative work should focus on a more detailed exploration of animal pattern perception abilities at the levels of sequence, rhythm and stress.

Notable accomplishments of the grant include the development of several new and innovative paradigms for cognitive testing, the creation of an “open source” physical and software infrastructure for designing and running cognitive experiments with humans and animals, and two successful international scientific meetings. All computer code produced is open source, and freely available to the entire scientific community (see www.somacca.org). We have demonstrated the value of rigorous theoretical framework based on the theory of computation both in reviews and experiments, have introduced several novel artificial grammars, and tested them in animals and humans. We have broken new ground in empirical aesthetics, introducing and extensively testing a methodology allowing virtually anyone to produce interesting and attractive visual stimuli (“FlexTiles”). Finally, a large number of young scientists have received training as part of the grant, including 5 PhD student and 8 post-docs.