Natural language is a uniquely convenient means of communication due to, among its other properties, its flexibility and its openness to interpretation. These properties of natural language are largely made possible by its heavy dependence on context and common ground. Drawing on elements of Clark’s account of language use, we view natural language interactions as a coordination problem involving agents who work together to convey and thus coordinate their interaction goals. In the modeling work presented here, a sequence of interrelated modules developed in the Polyscheme cognitive architecture is used to implement several stages of reasoning the user of a simple video application would expect an addressee—ultimately, the application—to work through, if the interaction goal was to locate a scene they had previously viewed together.

Modal verbs occur in contexts which convey information about non-actual states of affairs as well as in contexts which convey information about the actual world of the discourse. Modeling the semantic interpretation of non-actual states of affairs is notoriously complicated, sometimes requiring modal logic, belief revision, non-monotonic reasoning, and multi-agent autoepistemic models. This work presents linguistic features which disambiguate those instances of the past tense modal verbs `could’ and `would’ which occur in contexts where the proposition in the scope of the modal is not true in the actual world of the discourse from those instances which presuppose or entail that an event in their scope occurred in the actual world of the discourse. It also illustrates the complexity of the role of modal verbs in semantic interpretation and, consequently, the limitations of state of the art inference systems with respect to modal verbs. The features suggested for improving modal verb interpretation are based on the analysis of corpus data and insights from the linguistic literature.

Situated dialogue is prominent in the robot navigation task, where a human gives route instructions (i.e., a sequence of navigation commands) to an agent. We propose an approach for situated dialogue agents whereby they use strategies such as asking questions to repair or recover from unclear instructions, namely those that an agent misunderstands or considers ambiguous. Most immediately in this work we study examples from existing human-human dialogue corpora and relate them to our proposed approach.

We present a graph-theoretic model of discourse based on the Questions Under Discussion (QUD) framework. Questions and assertions are treated as edges connecting discourse states in a rooted graph, modeling the introduction and resolution of various QUDs as paths through this graph. The amount of common ground presupposed by interlocutors at any given point in a discourse corresponds to graphical depth. We introduce a new task-oriented dialogue corpus and show that experts, presuming a richer common ground, initiate discourse at a deeper level than novices. The QUD-graph model thus enables us to quantify the experthood of a speaker relative to a fixed domain and to characterize the ways in which rich common ground facilitates more efficient communication.

The concept of common ground — the mutual understanding of context and conventions — is central to philosophical accounts of mendacity; its use is to determine the meaning of linguistic expressions and the significance of physical acts, and to distinguish certain statements as conveying a conventional promise, warranty, or expectation of sincerity. Lying necessarily involves an abuse of common ground, namely the willful violation of conventions regulating sincerity. The ‘lying machine’ is an AI system that purposely abuses common ground as an effective means for practicing mendacity and lesser deceptions. The machine's method is to conceive and articulate sophisms — perversions of normative reason and communication — crafted to subvert its audience's beliefs. Elements of this paper (i) explain the described use of common ground in philosophical accounts of mendacity, (ii) motivate arguments and illusions as stratagem for deception, (iii) encapsulate the lying machine's design and operation, and (iv) summarize human-subject experiments that confirm the lying machine's arguments are, in fact, deceptive.

I ask why humans are smarter than other primates, and I hypothesize that an important part of the answer lies in what I call the Strong Story Hypothesis, which holds that story telling and understanding have a central role in human intelligence. Next, I introduce another hypothesis, the Driven Perception Hypothesis, which holds that we derive much of our commonsense, including the commonsense required in story understanding, by deploying our perceptual apparatus on real and imagined events. Then, after discussing methodology, I describe the representations and methods embodied in the Genesis system, a story-understanding system that analyzes stories ranging from precis of Shakespeare's plots to descriptions of conflicts in cyberspace. The Genesis system works with short story summaries, provided in English, together with low-level commonsense rules and higher-level reflection patterns, likewise expressed in English. Using only a small collection of commonsense rules and reflection patterns, Genesis demonstrates several story-understanding capabilities, such as determining that both Macbeth and the 2007 Russia-Estonia Cyberwar involve revenge, even though neither the word revenge nor any of its synonyms are mentioned. Finally, I describe Rao's Visio-Spatial Reasoning System, a system that recognizes activities such as approaching, jumping, and giving, and answers commonsense questions posed by Genesis.

The ViewGen model of belief ascription assumes that each agent involved in a conversation has a belief space which includes models of what other parties to the conversation believe. The distinctive notion is that a basic procedure, called belief ascription, allows belief spaces to be amalgamated so as to model the updating and augmentation of belief environments. In this paper we extend the ViewGen model to a more general account of reference phenomena, in particular by the notion of a reachable ascription set (RAS) that links intensional objects across belief environments so as to locate the most heuristically plausible referent at a given point in a conversation. The key notion is the location and attachment of entities that may be under different descriptions, the consequent updating of the system's beliefs about other agents by default, and the role in that process of a speaker's and hearer's protocols that ensure that the choice is the appropriate one. An important characteristic of this model is that each communicator considers nothing beyond his own belief space. A conclusion we shall draw is that traditional binary distinctions in this area (like de dicto/de re and attributive/referential) neither classify the examples effectively nor do they assist in locating referents, whereas the single procedure we suggest does both. We also suggest ways in which this analysis can also illuminate other traditional distinctions such as referential and attributive use. The description here is not on an implemented system with results but a theoretical tool to be implemented within an established dialogue platform (such as Wilks et al. 2011).

Inference techniques play a central role in many cognitive systems. They transform low-level observations of the environment into high-level, actionable knowledge which then gets used by mechanisms that drive action, problem-solving, and learning. This paper presents an initial effort at combining results from AI and psychology into a pragmatic and scalable computational reasoning system. Our approach combines a numeric notion of plausibility with first-order logic to produce an incremental inference engine that is guided by heuristics derived from the psychological literature. We illustrate core ideas with detailed examples and discuss the advantages of the approach with respect to cognitive systems.

This paper proposes a novel agenda for cognitive systems research focused on the "social agency" problem, which concerns acting to produce mental states in other agents in addition to physical states of the world. The capacity for social agency will enable agents to perform a wide array of tasks in close association with people and is a valuable first step towards broader social cognition. We argue that existing cognitive systems have not addressed social agency because they lack a number of the required mechanisms. We describe an initial approach set in a toy scenario based on capabilities native to the ICARUS cognitive architecture. We utilize an analysis of this approach to highlight the open issues required for social agency and to encourage other researchers to address this important problem.

Cognitive systems with human-level intelligence must dis­play a wide range of abilities, including reasoning about the beliefs of others, hypothetical and future situations, quanti­fiers, probabilities, and counterfactuals. While each of these deals in some way with reasoning about alternative states of reality, no single knowledge representation framework deals with them in a unified and scalable manner. As a conse­quence it is difficult to build cognitive systems for domains that require each of these abilities to be used together. To enable this integration we propose a representational framework based on synchronizing beliefs between worlds. Using this framework, each of these tasks can be reformu­lated into a reasoning problem involving worlds. This demonstrates that the notions of worlds and inheritance can bring significant parsimony and broad new abilities to knowledge representation.