We present a new theory of categorization based on an information-theoretic rational analysis. To evaluate this theory, we investigate how well it can account for key findings from classic categorization experiments conducted by Hayes-Roth and Hayes-Roth (1977), Medin and Schaffer (1978), and Smith and Minda (1998). We find that it explains the human categorization behavior as well as (or better) than the independent cue and context models (Medin Schaffer, 1978), the rational model of categorization (Anderson, 1991), and a hierarchical Dirichlet process model (Griffiths et al., 2007).

In this paper, I review approaches for acquiring hierarchical knowledge to improve the effectiveness of planning systems. First I note some benefits of such hierarchical content and the advantages of learning over manual construction. After this, I consider alternative paradigms for encoding and acquiring plan expertise before turning to hierarchical task networks. I specify the inputs to HTN learners and three subproblems they must address: identifying hierarchical structure, unifying method heads, and finding method conditions. Finally, I pose seven challenges the community should pursue so that techniques for learning HTNs can reach their full potential.

Open-world AI is characterized by sudden novel changes in a domain that are outside the scope of the training data, or the deployment of an agent in conditions that violate the implicit or explicit assumptions of the designer. In such situations, the AI system must detect the novelty and adapt in a short time frame. In this introduction to the special issue on open-world AI, we discuss the background and motivation for this new research area and define the field in the context of similar AI challenges. We then discuss recent research in the area that has made significant contributions to the field. Many of those contributions are reflected in the papers of this special issue, which we summarize alongside more traditional approaches to open-world AI. Finally, we discuss future directions for the field.

In this paper, we review PUG, a cognitive architecture for embodied agents, and report extensions that let it represent and reason about spatial relations. The framework posits graded concepts that are grounded in perception and integrates symbolic reasoning with continuous control. After describing the architecture, we discuss how an extended version supports places, encoded as virtual objects defined by distances to reference entities, and reasons about them as if they were visible. We demonstrate PUG's control of a simulated robot that approaches targets and avoids obstacles in a simple two-dimensional environment. In closing, we discuss related research on agent architectures, robotic control, and spatial cognition, along with our plans to extend the framework's capabilities for spatial representation and reasoning.

This paper poses the challenge of developing and evaluating integrated systems for computational scientific discovery. We note some distinguishing characteristics of discovery tasks, examine eight component abilities, review previous successes at partial integration, and consider hurdles the AI research community must leap to transform the vision for integrated discovery into reality. In closing, we discuss promising scientific domains in which to test such computational artifacts.

In this paper, we review hierarchical problem networks, which encode knowledge about how to decompose planning tasks, and report an approach to learning this expertise from sample solutions. In this framework, procedural knowledge comprises a set of conditional methods that decompose problems – sets of goals – into subproblems. Problem solving involves search through a space of hierarchical plans that achieve top-level goals. Acquisition involves creation of new methods, including state conditions for when they are relevant and goal conditions for when to avoid them. We describe HPNL, a system that learns new methods by analyzing sample hierarchical plans, using violated constraints to identify state conditions and ordering conflicts to determine goal conditions. Experiments with on-line learning in three planning domains demonstrate that HPNL acquires expertise that reduces search on novel problems and examine the importance of learning goal conditions. In closing, we contrast the approach with earlier methods for acquiring search-control knowledge, including explanation-based learning and inductive logic programming. We also discuss limitations and plans for future research.

In this paper, I pose a major challenge for AI researchers: to develop systems that learn in a human-like manner. I briefly review the history of machine learning, noting that early work made close contact with results from cognitive psychology but that this is no longer the case. I identify seven characteristics of human behavior that, if reproduced, would offer better ways to acquire expertise than statistical induction over massive training sets. I illustrate these points with two domains - mathematics and driving - where people are effective learners and review systems that address them. In closing, I suggest ways to encourage more research on human-like learning.

In this paper, we review the PUG cognitive architecture for embodied agents, which posits grounded and graded concepts, integration of symbolic planning with continuous control, and mental simulation to evaluate candidate trajectories. We also describe extensions that will let the framework represent spatial knowledge about places and use this content to move an agent through its environment.

operators from experience. Research on learning action models (e.g., Wang, 1995) is relevant here, but it has seldom been applied to constructing problem spaces. In summary, there are promising avenues for extending the heuristic search theory in this direction, but it merits more attention than given to date. A second drawback concerns when to create and eliminate problems. People identify and pursue new top-level tasks; they also decide when problems have been solved to their satisfaction and when to abandon them. Classic work on problem solving offers ways to generate subproblems (e.g., Newell et al., 1960), but introducing top-level tasks is another matter. Research on ‘goal reasoning’ (Aha, Cox, & Muñoz-Avila, 2013). has addressed this issue, often by positing generic rules or ‘motives’ that produce concrete goals under specified conditions. These shift the focus of an agent’s attention to new problems that drive cognitive and physical behavior in different directions. The second topic, when to declare success, is a distinct issue that relates to Simon’s (1956) theory of satisficing, which states that people halt when they encounter an option that is ‘good enough’. Classic work on problem solving assumes that goal descriptions are matched in an all-or-none manner, but many settings diverge from this assumption. Using numeric evaluation functions to guide search lends itself to other termination schemes, such as halting when the value meets some threshold, which maps onto Simon’s notion of an aspiration level in satisficing. The concept has typically been applied to simple choice tasks, in which the decision maker selects one candidate from a set, but it is equally relevant to problem solving, where it can serve as a criterion for success. The literature includes only a few studies of this topic (e.g., Gajderowicz et al., 2018), so this is another area that needs more research.4 Simple accounts treat aspiration level as constant, but it can clearly vary over time, as when repeated failures lead a person to revise expectations downward or successes upward. Repeated failures can also cause the problem solver to abandon a task when he decides that expending more effort is unlikely to produce results. A third limitation involves the inability of heuristic search, at least on its own, to explain insight, in which someone repeatedly fails to solve a problem until they suddenly and unexpectedly ‘see’ the solution that evaded them before. Laboratory examples include the nine-dot puzzle and matchstick problems, which people find difficult to solve without hints, but insight has also been linked repeatedly to creativity in science and design. This phenomenon was a central concern of Gestalt theorists (Koffka, 1935; Kohler, 1940; Wertheimer, 1959), who posited that it involves restructuring the problem so that, perceived in a new way, the answer is obvious. Ohlsson (1984a) provides an excellent reconstruction of Gestalt theory and how it differs from Newell and Simon’s framework. They share the assumption that unsolved problems have ‘gaps’ that must be filled, but they propose distinct mechanisms. In Gestalt restructuring, a problem structure is subject to ‘forces’ that, when unbalanced, draw it inevitably toward a new structure that is in a dynamic equilibrium, much like a physical system. In contrast, Newell and Simon allowed for multiple paths in a problem space that heuristic search must explore. A few efforts have extended the latter framework to explain insight effects. Simon (1966) suggested that extensive search followed by selective forgetting could reveal a previously obscured solution, whereas Jones and Langley (2005) reported an account that combined means-ends analysis with spreading activation retrieval. Ohlsson (1984b) proposed search through an augmented problem space that included operators for altering representations. In the AI literature, Amarel (1968) noted the influence of representations on problem solving and, more recently, Riddle, Barley, and Franco (2013) have analyzed meta-level operators that revise them. However, work on this topic has been limited and it deserves much more attention from the research community. Work on partial satisfaction planning (Benton, Do, & Kambhampati, 2009), which finds solutions that achieve a subset of target goals, is related but typically insists on optimality, which runs directly counter to Simon’s definition.

This paper presents a novel approach to the acquisition of language models from corpora. The framework builds on Cobweb, an early system for constructing taxonomic hierarchies of probabilistic concepts that used a tabular, attribute-value encoding of training cases and concepts, making it unsuitable for sequential input like language. In response, we explore three new extensions to Cobweb -- the Word, Leaf, and Path variants. These systems encode each training case as an anchor word and surrounding context words, and they store probabilistic descriptions of concepts as distributions over anchor and context information. As in the original Cobweb, a performance element sorts a new instance downward through the hierarchy and uses the final node to predict missing features. Learning is interleaved with performance, updating concept probabilities and hierarchy structure as classification occurs. Thus, the new approaches process training cases in an incremental, online manner that it very different from most methods for statistical language learning. We examine how well the three variants place synonyms together and keep homonyms apart, their ability to recall synonyms as a function of training set size, and their training efficiency. Finally, we discuss related work on incremental learning and directions for further research.

In this paper we examine the behavior of a human-computer system for crisis response. As one instance of crisis management, we describe the task of responding to spills and fires involving hazardous materials. We then describe Inca, an intelligent assistant for planning and scheduling in this domain, and its relation to human users. We focus on Inca’s strategy of retrieving a case from a case library, seeding the initial schedule, and then helping the user adapt this seed. We also present three hypotheses about the behavior of this mixed-initiative system and some experiments designed to test them. The results suggest that our approach leads to faster response development than user-generated or automatically-generated schedules but without sacrificing solution quality.

This paper briefly proposes a theory of emotions that clarifies their role in architectures for intelligent agents. The account posits that emotions take the form of symbolic cognitive structures, that generic emotional rules produce concrete instances of these relational concepts, that such rules underlie both generation and understanding of emotions, and that their results play a metacognitive role in influencing behavior. In this framework, emotions are central to high-level, human-like cognition.

In this paper, I examine challenges that arise in developing theories of open-world learning. After defining the problem, I review some theories from the history of chemistry, biology, geology, and AI, along with the importance of inductive bias to constrain the learning process. Classic cognitive architectures offer one source for such guidance, but their generality provides little aid on this front. Instead, I propose that more constrained architectures for embodied agents have greater potential, as they make commitments about the form of domain knowledge used to describe environments, as well as the processes that operate over them. In addition, I hypothesize that autonomous agents must include motivational structures that drive behavior and that changes to the environment can lead to their revision as well. I argue that a full account of open-world learning should make commitments about the structures and processes that underlie these capabilities. Autonomous Agency in Open Worlds Advances in sensors, effectors, memories, and processors have led to autonomous agents that are far more capable and common than those from only a decade ago. These take on many different forms, from self-driving cars and delivery drones to military robots and planetary rovers. The development of such systems typically relies on collection and processing of very large training sets to create accurate pattern recognizers and efficient controllers. This approach is viable for some applications and certain contexts, but it depends on two related assumptions: the environment will not change in important ways; and the agent’s expertise will remain accurate and appropriate. Unfortunately, these postulates will not hold in many real-world settings. We would like autonomous agents that are robust to such shifts. For example, consider an unmanned aerial drone on an exploratory mission in the Amazon rainforest. The system’s expertise remains accurate and its behavior is acceptable until an airborne spider’s web tangles one of its rotor blades, a large predatory bird attacks it from above, a strong updraft pulls it off course, a dense fog bank degrades its visibility, or high humidity causes intermittent shorts in its camera controller. A truly flexible autonomous agent would reCopyright c © 2022, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. alize, in each case, that its expertise was outdated and adapt rapidly enough to still achieve the mission goals. Scenarios of this sort raise the challenge of open-world learning, a class of problems introduced by DARPA’s SAILON program (Senator, 2019). We can state this problem as: • Given: An agent architecture that operates in some class of tasks and environments; • Given: Expertise that supports acceptable performance for these tasks and environments; • Given: Limited experience after sudden, unannounced changes to the environment degrade performance; • Find: When the environmental change occurs and what revised expertise will give acceptable performance. This formulation applies to many agents, environments, and tasks, regardless of whether their initial expertise is handcrafted or learned from experience. The problem of open-world learning addresses the very heart of what we mean by the term ‘autonomous agent’. We say that an entity is an ‘agent’ if it carries out actions that affect its environment over time. However, in this light, teleoperated robots and remote-controlled drones would count as agents. We say that an agent is ‘autonomous’ if it operates independently and without supervision, but this can hold to different degrees. Thermostats are autonomous but only in a very narrow context. Robot vacuum cleaners have a broader range of behaviors but have been programmed by humans. Humans fall at the spectrum’s extreme end, because they can not only adapt their model of the world but, in some cases, alter their own motivations and value systems. Some readers may question why open-world learning poses a challenge, in that modern techniques for machine learning have been widely advertised as the solution to nearly any problem. However, remember that environmental shifts can be sudden and unannounced and that the agent must detect them and repair its expertise rapidly. The most widely adopted methods for classification learning, despite their success in some settings, rely on batch processing and require many labeled training cases, neither of which are sufficient here. Reinforcement learning, a popular approach to sequential action selection, typically requires many runs on a simulator, which will not be available for unfamiliar physical environments. In summary, mainstream approaches to machine learning are ill suited to such scenarios. In the pages that follow, I discuss a number of issues related to theories of open-world learning. First, I review some familiar examples from the history of science, including artificial intelligence, and their lessons for accounts of learning. After noting the need for some form of inductive bias to make the search process tractable, I consider what cognitive architectures can offer to this end. Upon concluding that they offer only weak constraints, I turn to more specialized architectures for embodied agents that include theoretical commitments about knowledge of the physical environment. Finally, I argue that a complete theory of open-world learning must address not only how agents can alter their environmental models, but how they can alter their own motivations in response to such changes. Theories of Open-World Learning Given that we want the research community to develop theories of open-world learning, we should consider the form that such accounts might take. The history of science, including the early phases of artificial intelligence, offer compelling examples from which researchers can draw useful lessons. In this section, I review three classic theories from the disciplines of chemistry, biology, and geology, along with three others cases from the study of intelligent systems. In closing, I revisit some familiar ideas about the need for constraints in machine learning. Scientific theories aim to explain observed phenomena in terms of a set of interconnected claims or assumptions. Here are three well-known examples: • Dalton’s (1808) atomic theory posited that macroscopic objects are made from tiny molecules, each involving atoms of nondecomposable elements. Moreover, chemical reactions transform some types of molecules into other types by rearranging their constituent atoms. • Pasteur’s (1880) germ theory of disease proposed that many illnesses are caused by small organisms that invade the body and attack it. In addition, these germs spread from one host to another through the process of infection. • Hess’s (1962) theory of plate tectonics stated that interlocking plates make up the Earth’s surface, with mountains and deep sea trenches at their interfaces. These plates move very slowly under, over, and against each other to produce large-scale geological formations. There are many analogous examples from the history of science, but we can draw some tentative conclusions from this set about the character of theories. Despite their many differences in form and content, scientific theories nearly always include postulates that impose qualitative constraints on the domain under study. This holds even when the account also includes quantitative elements, which are often introduced after a field has agreed about their qualitative aspects. Moreover, theories posit both structures (e.g., entities and their relations) and processes that operate over and transform them. For instance, molecules and atoms are structures in the atomic theory, whereas chemical reactions are processes that affect them. Also, theories are abstract enough that they cannot be tested directly; this can only occur when one has added enough assumptions to produce operational models. For example, the atomic theory must be augmented by specific claims about the constituents of particular molecules, while germ theory requires associations between specific microorganisms and diseases. Finally, scientific theories regularly elaborate earlier ones with which they share assumptions, as the immune theory builds on the more basic germ theory. We should also consider examples from the early days of artificial intelligence, which illustrate many of the same characteristics. These include three classic theories: • Physical symbol systems (Newell & Simon, 1976), which posits that mental structures consist of symbols (persistent physical patterns) and symbol structures (organized sets of such symbols), which in turn can designate other entities or activities. This theory also proposes mental processes that create, modify, and interpret an evolving sequence of these symbol structures. • Production systems (Newell, 1966), which elaborates on the first theory by postulating memories that contain sets of modular elements encoded as symbol structures, including a rapidly changing working memory and a more stable long-term store with condition-action rules. Processing involves repeatedly matching rules against elements in working memory and using them to alter its contents, which in turn enables new matches. • Heuristic search (Newell & Simon, 1976), which also extends the first framework by declaring that problem solving relies on symbol structures to denote candidate solutions, generators of candidates, criteria for acceptance, and heuristics. This theory assumes processes for generating candidate solutions, testing them for acceptability, and using heuristics to guide choices. These examples clarify that theories of intelligent behavior specify both structures and processes that operate over them. They also have the same abstract, qualitative character as the cases from chemistry, biology, and geology. These observations are relevant for researchers wh

Autonomous agents have many applications in familiar situations, but they also have great potential to help us understand novel settings. In this paper, I propose a new challenge for the AI research community: developing embodied systems that not only explore new environments but also characterize them in scientific terms. Illustrative examples include autonomous rovers on planetary surfaces and unmanned vehicles on undersea missions. I review two relevant paradigms: robotic agents that explore unknown areas and computational systems that discover scientific models. In each case, I specify the problem, identify component functions, describe current abilities, and note remaining limitations. Finally, I discuss obstacles that the community must overcome before it can develop integrated agents of exploration and discovery.

This paper presents a theory that unifies the frameworks of symbolic problem solving and continuous control at a deeper level than previously attempted. The approach builds on the earlier PUG architecture, but it also incorporates some important new ideas: symbolic concepts are grounded in numeric attributes and can match to different degrees; symbolic skills include equations for control attributes that are modulated by mismatches of target concepts; and reactive execution uses these skills to compute control values based on mismatches. We describe C, an implemented version of this theory and demonstrate its behavior in a simulated continuous robot environment. We conclude by discussing links to earlier work and directions for future efforts.

In this essay, I discuss the importance of explanation to the computational study of cognitive systems. I distinguish between two common senses of the term – accounts of situations or events and the processes that produce them – as well as two distinct activities – constructing such accounts and communicating them to others. After this, I examine some representational issues that arise in both contexts, including the central role of knowledge and the varieties of explanatory structures. Next I consider in more detail the mechanisms that underlie the generation of explanations and the component abilities that support their communication. Finally, I conclude by noting some research challenges that cognitive systems researchers should pursue to further the field’s understanding of these important mental faculties.

In this essay, I discuss the role of communication in scientific research and its relevance to publications on cognitive systems. In particular, I discuss the content that our field’s papers should cover in order to convey their contributions to readers. This material should include a clear statement of the cognitive task under study and the target behaviors or phenomena one wants to reproduce. Articles should also present a high-level theory of these behaviors and distinguish it from a more detailed system or model that instantiates the theory. In addition, authors should make explicit behavioral claims and support them with evidence, empirical or analytical. Other topics should include related research, limits of the approach described, and plans to remedy the latter in future work. 1. The Importance of Scientific Communication Science is a communal enterprise. The image of the solitary scientist who works in isolation is a misleading myth. Even researchers who do not participate in explicit teams spend time tracking others’ progress and incorporating selected insights into their own efforts. Communication is central to this process, as it lets scientists exchange ideas, identify points of agreement and disagreement, and build on earlier results. Newton (1675) noted the importance of previous work in physics, writing “If I have seen further it is by standing on the shoulders of Giants”. However, this dependence holds equally well for other natural sciences and even for ‘sciences of the artificial’ (Simon, 1969), which includes cognitive systems. Such cumulative progress is only possible through effective communication among researchers. Scientists communicate in many different ways – in conference talks, during informal breaks, over the telephone, and through electronic mail – but the classic medium is published articles. This makes the art of writing such papers a central element of success in any field. As Winston (in press) discusses, good writing depends on many factors, some of them independent of a discipline’s content, and I encourage readers to follow his sage advice.1 However, each scientific field also has special concerns, and cognitive systems is distinct enough that it merits its own style of research paper. Unlike mainstream AI, our discipline does not focus on narrowly defined technical schemes that are familiar to specific subcommunities (e.g., statistical learning or automated planning). Rather, the broad scope and inherent diversity of cognitive systems means that articles must provide substantial background and state their ideas and their contributions clearly and explicitly. 1. Winston also discusses communicating ideas through formal talks and lectures, which raise many similar challenges. c © 2019 Cognitive Systems Foundation. All rights reserved.

In this paper, I pose a new research challenge – to develop intelligent agents that exhibit radical autonomy by responding to sudden, long-term changes in their environments. I illustrate this idea with examples, identify abilities that support it, and argue that, although each ability has been studied in isolation, they have not been combined into integrated systems. In addition, I propose a framework for characterizing environments in which goal-directed physical agents operate, along with specifying the ways in which those environments can change over time. In closing, I outline some approaches to the empirical study of such open-world learning.

Humans regularly explain observations of their environment in terms of background knowledge. This process is better characterized as abduction than as deduction, since it often requires introduction of assumptions about unobserved relations. In this paper, we present a theory of abductive explanation that builds on earlier work but extends it in new directions. The theory distinguishes between definitions and constraints, with the former used to elaborate existing explanations and the latter used to detect and repair inconsistencies. We also describe PENUMBRA, an implemented system that instantiates the theory, and demonstrate its behavior on a number of domains. The system carries out heuristic search through a space of explanations, processing observations incrementally, and generating alternative accounts for a given set of inputs. We conclude by discussing related approaches to explanation, limitations of the implementation, and directions for future research. 1. Background and Motivation One distinctive feature of human cognition is the ability to understand complex situations and events. This invariably involves explaining observations in terms of available knowledge. Moreover, these explanations are typically abductive in character, in that they incorporate plausible assumptions that are neither observed nor derived deductively. Abductive explanation is a general ability that arises in many contexts, from sentence processing and story understanding (Winston, 2012) to scene interpretation and plan recognition (Blaylock & Allen, 2005) to diagnosis (Reggia, Nau, & Wang, 1985). We would like a computational theory of the structures that underlie this capacity and the processes that operate over them. Ultimately, this should contribute to a more comprehensive cognitive architecture (Langley, Laird, & Rogers, 2009) that supports goal-directed activity over time, but here we will focus only on conceptual understanding. Let us consider a simple example. Suppose we are told that Abe possesses some cash and Bob possesses a car, but that later Abe possesses the same car. Although we did not observe any transaction, we can reasonably assume that one took place. Two explanations come immediately to mind. One is that Abe bought the car from Bob using money; another is that Abe stole the car from Bob by threatening him in some way. We also know these two explanations are mutually c © 2019 Cognitive Systems Foundation. All rights reserved. P. LANGLEY AND B. MEADOWS exclusive, in that purchases and robbery are two distinct ways to transfer possession of objects. This means that we must not only introduce plausible assumptions about unobserved events, but consider the competing explanations and keep them separate. Later, we may hear that Abe actually gave money to Bob, eliminating theft as an alternative. More complex examples would involve multi-step inference chains that generate hierarchical accounts of observations. In this paper, we present a cognitive systems account of such abductive explanations. Our analysis draws on standard ideas from the paradigm, including a focus on high-level cognition, the importance of structured representations and knowledge, a reliance on heuristic search, and incorporation of constraints from human behavior, such as incremental processing of observations. The approach that we describe builds directly on two earlier efforts (Bridewell & Langley, 2011; Meadows, Langley, & Emery, 2014) in this area, but extends them to incorporate richer representations and novel reasoning mechanisms. In the next section, we discuss two formulations of the abductive explanation task, along with prior results in each framework, and clarify our reasons for selecting one of them. After this, we describe a new theory for this ability, focusing first on assumptions about cognitive structures and then on processes that inspect and manipulate them. Next we report PENUMBRA, an implemented system that instantiates these theoretical ideas, along with its behavior on multiple scenarios that demonstrate its coverage. We close by discussing links to earlier work, noting limits of the implementation, and proposing directions for additional research in this area. 2. Two Formulations of Abductive Explanation We can define any cognitive task in terms of the information provided as inputs and the content generated as outputs. However, there are often different ways to translate an informal problem into a formal specification. The literature on abductive explanation has explored two distinct statements of this mental task that we should discuss before proceeding further. These treatments are orthogonal to whether inputs are processed incrementally, an important feature of human processing whose discussion we will delay until a later section. The first formulation borrows from classic treatments of abduction in logic and the philosophy of science (Peirce, 1878; Hempel, 1966). We can state it as: • Given: A set of general knowledge elements K (e.g., relational rules) • Given: A set of specific observed facts O (e.g., relational literals) • Find: A set of specific plausible assumptions A (e.g., relational literals) • Find: A set of proof trees P that derive elements of O from A and other elements of O with K The key idea here is that the resulting explanation, a set of linked proof trees, must contain a proof for each observed fact. These may include default assumptions as terminal nodes, which can be shared across different proof trees, but each observation must follow deductively from these assumptions and from other observations by reasoning over available knowledge. Proof trees may correspond to causal chains, as in many scientific explanations, but this is not a requirement. We will refer to this formulation as derivational abduction because observations must be derived from other beliefs. This paradigm has received considerable attention in the AI community. For example, Reggia et al. (1985) adopted the approach in their work on diagnosis, which inferred unobserved diseases that caused observed symptoms, and Hobbs et al. (1993) used it in their ap-

In this chapter, we review research on computational approaches to scientific discovery, starting with early work on the induction of numeric laws before turning to the construction of models that explain observations in terms of domain knowledge. We focus especially on inductive process modeling, which involves finding a set of linked differential equations, organized into processes, that reproduce, predict, and explain multivariate time series. We review the notion of quantitative process models, present two approaches to their construction that search through a space of model structures and associated parameters, and report their successful application to the explanation of ecological data. After this, we explore the relevance of process models to the social sciences, including the reasons they seem appropriate and some challenges to discovering them. In closing, we discuss other causal frameworks, including structural equation models and agent-based accounts, that researchers have developed to construct models of social phenomena.