COMPS: Conceptual Minimal Pair Sentences for testing Robust Property Knowledge and its Inheritance in Pre-trained Language Models

Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics, pages 2928–2949 May 2-6, 2023 ©2023 Association for Computational Linguistics COMPS: Conceptual Minimal Pair Sentences for testing Robust Property Knowledge and its Inheritance in Pre-trained Language Models Kanishka Misra Purdue University kmisra@purdue.edu Julia Rayz Purdue University jtaylor1@purdue.edu Allyson Ettinger University of Chicago aettinger@uchicago.edu Abstract A characteristic feature of human semantic cog- nition is its ability to not only store and retrieve the properties of concepts observed through experience, but to also facilitate the inheri- tance of properties (can breathe) from superor- dinate concepts (ANIMAL) to their subordinates (DOG)—i.e. demonstrate property inheritance. In this paper, we present COMPS, a collection of English minimal pair sentences that jointly tests pre-trained language models (PLMs) on their ability to attribute properties to concepts and their ability to demonstrate property inheri- tance behavior. Analyses of 22 different PLMs on COMPS reveal that they can easily distin- guish between concepts on the basis of a prop- erty when they are trivially different, but find it relatively difficult when concepts are related on the basis of nuanced knowledge represen- tations. Furthermore, we find that PLMs can show behaviors suggesting successful property inheritance in simple contexts, but fail in the presence of distracting information, which de- creases the performance of many models some- times even below chance. This lack of robust- ness in demonstrating simple reasoning raises important questions about PLMs’ capacity to make correct inferences even when they appear to possess the prerequisite knowledge. 1 Introduction The ability to learn, update and deploy one’s knowl- edge about concepts (ROBIN, CHAIR) and their properties (can fly, can be sat on), observed dur- ing everyday experience is fundamental to human semantic cognition (Murphy, 2002; Rogers and Mc- Clelland, 2004; Rips et al., 2012). Knowledge of a concept’s properties, combined with the ability to infer the IsA relation (Sloman, 1998; Murphy, 2003) leads to an important behavior known as property inheritance (Quillian, 1967; Smith and Estes, 1978; Murphy, 2002), where subordinates of a concept inherit its properties. For instance, one is likely to infer that an entity called luna can meow, has a tail, is a mammal, etc., even if all they know is that it is a cat. The close connection between a word’s meaning and its conceptual repre- sentation makes these abilities crucial to language understanding (Murphy, 2002; Lake and Murphy, 2021), making it critical for computational mod- els of language processing to also exhibit behav- ior consistent with these capacities. Indeed, mod- ern pre-trained language models (PLMs; Devlin et al., 2019; Brown et al., 2020, etc.) have made impressive empirical strides in eliciting general knowledge about real world concepts and entities (Petroni et al., 2019; Weir et al., 2020, i.a.), as well as in demonstrating isomorphism with real world abstractions like direction and color (Abdou et al., 2021; Patel and Pavlick, 2022), often times without even having been explicitly trained to do so. At the same time, their ability to robustly demonstrate …

Content truncated to 500 words