We consider the possible role of current large language models (LLMs) in the study of human linguistic cognition. We focus on the use of such models as proxies for theories of cognition that are relatively linguistically-neutral in their representations and learning but differ from current LLMs in key ways. We illustrate this potential use of LLMs as proxies for theories of cognition in the context of two kinds of questions: (a) whether the target theory accounts for the acquisition of a given pattern from a given corpus; and (b) whether the target theory makes a given typologically-attested pattern easier to acquire than another, typologically-unattested pattern. For each of the two questions we show, building on recent literature, how current LLMs can potentially be of help, but we note that at present this help is quite limited.

While Large Language Models (LLMs) have found success in real-world applications, their underlying explanatory process is still poorly understood. This paper proposes IBE-Eval, a framework inspired by philosophical accounts on Inference to the Best Explanation (IBE) to advance the interpretation and evaluation of LLMs' explanations. IBE-Eval estimates the plausibility of natural language explanations through a combination of explicit logical and linguistic features including: consistency, parsimony, coherence, and uncertainty. Extensive experiments are conducted on Causal Question Answering (CQA), where \textit{IBE-Eval} is tasked to select the most plausible causal explanation amongst competing ones generated by LLMs (i.e., GPT 3.5 and Llama 2). The experiments reveal that IBE-Eval can successfully identify the best explanation with up to 77\% accuracy ($\approx 27\%$ above random), improving upon a GPT 3.5-as-a-Judge baseline ($\approx+17\%$) while being intrinsically more efficient and interpretable. Additional analyses suggest that, despite model-specific variances, LLM-generated explanations tend to conform to IBE criteria and that IBE-Eval is significantly correlated with human judgment, opening up opportunities for future development of automated explanation verification tools.

The present paper looks at one of the most thorough articles on the intelligence of GPT, research conducted by engineers at Microsoft. Although there is a great deal of value in their work, I will argue that, for familiar philosophical reasons, their methodology, !Blackbox Interpretability"#is wrongheaded. But there is a better way. There is an exciting and emerging discipline of !Inner Interpretability"#(and specifically Mechanistic Interpretability) that aims to uncover the internal activations and weights of models in order to understand what they represent and the algorithms they implement. In my view, a crucial mistake in Black-box Interpretability is the failure to appreciate that how processes are carried out matters when it comes to intelligence and understanding. I can#t pretend to have a full story that provides both necessary and sufficient conditions for being intelligent, but I do think that Inner Interpretability dovetails nicely with plausible philosophical views of what intelligence requires. So the conclusion is modest, but the important point in my view is seeing how to get the research on the right track. Towards the end of the paper, I will show how some of the philosophical concepts can be used to further refine how Inner Interpretability is approached, so the paper helps draw out a profitable, future two-way exchange between Philosophers and Computer Scientists.

Motivating explanations in Bayesian networks using MAP-independence Johan Kwisthout We introduce MAP-independence as a novel concept in Bayesian networks, indicating potential impact of an intermediate (hidden) variable to the MAP explanation. We discuss how this concept may contribute to justifying MAP explanations, for example in the context of a decision support system. Abstract In decision support systems the motivation and justification of the system's diagnosis or classification is crucial for the acceptance of the system by the human user. In Bayesian networks a diagnosis or classification is typically formalized as the computation of the most probable joint value assignment to the hypothesis variables, given the observed values of the evidence variables (generally known as the MAP problem). While solving the MAP problem gives the most probable explanation of the evidence, the computation is a black box as far as the human user is concerned and it does not give additional insights that allow the user to appreciate and accept the decision. For example, a user might want to know to whether an unobserved variable could potentially (upon observation) impact the explanation, or whether it is irrelevant in this aspect.

Current discussions of "Explainable AI" (XAI) do not much consider the role of abduction in explanatory reasoning (see Mueller, et al., 2018). It might be worthwhile to pursue this, to develop intelligent systems that allow for the observation and analysis of abductive reasoning and the assessment of abductive reasoning as a learnable skill. Abductive inference has been defined in many ways. For example, it has been defined as the achievement of insight. Most often abduction is taken as a single, punctuated act of syllogistic reasoning, like making a deductive or inductive inference from given premises. In contrast, the originator of the concept of abduction---the American scientist/philosopher Charles Sanders Peirce---regarded abduction as an exploratory activity. In this regard, Peirce's insights about reasoning align with conclusions from modern psychological research. Since abduction is often defined as "inferring the best explanation," the challenge of implementing abductive reasoning and the challenge of automating the explanation process are closely linked. We explore these linkages in this report. This analysis provides a theoretical framework for understanding what the XAI researchers are already doing, it explains why some XAI projects are succeeding (or might succeed), and it leads to design advice.

Recent work in cognitive science has uncovered a diversity of explanatory values, or dimensions along which we judge explanations as better or worse. We propose a Bayesian account of how these values fit together to guide explanation. The resulting taxonomy provides a set of predictors for which explanations people prefer and shows how core values from psychology, statistics, and the philosophy of science emerge from a common mathematical framework. In addition to operationalizing the explanatory virtues associated with, for example, scientific argument-making, this framework also enables us to reinterpret the explanatory vices that drive conspiracy theories, delusions, and extremist ideologies. Intuitively, philosophically, and as seen in laboratory experiments, explanations are judged as better or worse on the basis of many different criteria. These explanatory values appear in early childhood [1, 2, 3, 4, 5] and their influence extends to some of the most sophisticated social knowledge formation processes we know [6]. We lack, however, an understanding of the origin of these values or an account of how they fit together to guide belief formation. The multiplicity of values also appears to conflict with Bayesian models of cognition, which speak solely in terms of degrees of beliefs and suggest we judge explanations as better or worse on the basis of a single quantity, the posterior likelihood (see Glossary). In this opinion, we show how to resolve these conflicts by arguing that previously-identified explanatory values capture different components of a full Bayesian calculation and, when considered together and weighed appropriately, implement Bayesian cognition. This framework shows how key explanatory values identified by laboratory experiments and philosophers of science--co-explanation, descriptiveness, precision, unification, power, and simplicity--emerge naturally from the mathematical structure of probabilistic inference, thereby reconciling them with Bayesian models of cognition [7, 8]. Second, it shows how these values combine to produce preferences for one explanation over another.

Finding the most probable explanation for observed variables in a Bayesian network is a notoriously intractable problem, particularly if there are hidden variables in the network. In this paper we examine the complexity of a related problem, that is, the problem of finding a set of sufficiently dissimilar, yet all plausible, explanations. Applications of this problem are, e.g., in search query results (you won't want 10 results that all link to the same website) or in decision support systems. We show that the problem of finding a 'good enough' explanation that differs in structure from the best explanation is at least as hard as finding the best explanation itself.

Every time a new tool or app is invented, a new word follows. So, let's tackle two that have been flying around our heads for the past few years: Machine Learning (ML) and Deep Learning (DL). Techies, business gurus, and marketers love these words and throw them around whether or not they understand the differences. Side Note: We know that this topic is old news, it's discussed continuously. Which is why we had to write about it, clearly it's not being fully understood because all the current content out there is either too simple or too complicated.

You happen to know that Tim and Harry have recently had a terrible row that ended their friendship. Now someone tells you that she just saw Tim and Harry jogging together. The best explanation for this that you can think of is that they made up. You conclude that they are friends again. One morning you enter the kitchen to find a plate and cup on the table, with breadcrumbs and a pat of butter on it, and surrounded by a jar of jam, a pack of sugar, and an empty carton of milk. You conclude that one of your house-mates got up at night to make him- or herself a midnight snack and was too tired to clear the table. This, you think, best explains the scene you are facing. To be sure, it might be that someone burgled the house and took the time to have a bite while on the job, or a house-mate might have arranged the things on the table without having a midnight snack but just to make you believe that someone had a midnight snack. But these hypotheses strike you as providing much more contrived explanations of the data than the one you infer to. Walking along the beach, you see what looks like a picture of Winston Churchill in the sand. It could be that, as in the opening pages of Hilary Putnam's (1981), what you see is actually the trace of an ant crawling on the beach. The much simpler, and therefore (you think) much better, explanation is that someone intentionally drew a picture of Churchill in the sand. That, in any case, is what you come away believing. In these examples, the conclusions do not follow logically from the premises.

Department of Computer Science Brown University Providence, RI 02912 Abstract In general, the best explanation for a given observation makes no promises on how good it is with respect to other alternative explanations. A major deficiency of message-passing schemes for belief revision in Bayesian networks is their inability to generate alternatives beyond the second best. In this paper, we present a general approach based on linear constraint systems that naturally generates alternative explanations in an orderly and highly efficient manner. This approach is then applied to cost-based abduction problems as well as belief revision in Bayesian networks. INTRODUCTION We are constantly faced with the problem of explaining the observations we have gathered with our senses. Our explanations are constructed by assuming certain facts or hypotheses which support our observations. For example, suppose I decide to phone my friend Tony at the office.