For intelligent systems to operate independently and make defensible judgments, they require efficient knowledge representation, reasoning, and planning strategies. This essay investigates these essential elements, going over their definitions, significance, methods, and incorporation in many applications. This study tries to provide insights into creating reliable and effective intelligent systems while resolving the difficulties they encounter by looking at recent developments and emerging trends.

We present a preliminary high-level formal theory, grounded on knowledge representation techniques and foundational ontologies, for the uniform and integrated representation of the different kinds of (quali- tative and quantitative) knowledge involved in the designing process. We discuss the conceptual nature of engineering design by individuating and analyzing the involved notions. These notions are then formally charac- terized by extending the DOLCE foundational ontology. Our ultimate purpose is twofold: (i) to contribute to foundational issues of design; (...) and (ii) to support the development of advanced modelling systems for (qualitative and quantitative) representation of design knowledge. (shrink)

Logic has been a—disputed—ingredient in the emergence and development of the now very large field known as knowledge representation and reasoning. In this book (in progress), I select some central topics in this highly fruitful, albeit controversial, association (e.g., non-monotonic reasoning, implicit belief, logical omniscience, closed world assumption), identifying their sources and analyzing/explaining their elaboration in highly influential published work.

This paper briefly outlines some advancements in paraconsistent logics for modelling knowledge representation and reasoning. Emphasis is given on the so-called Logics of Formal Inconsistency (LFIs), a class of paraconsistent logics that formally internalize the very concept(s) of consistency and inconsistency. A couple of specialized systems based on the LFIs will be reviewed, including belief revision and probabilistic reasoning. Potential applications of those systems in the AI area of KRR are tackled by illustrating some examples that emphasizes the (...) importance of a fine-tuned treatment of consistency in modelling reputation systems, preferences, argumentation, and evidence. (shrink)

According to the PubMed resource from the U.S. National Library of Medicine, over 750,000 scientific articles have been published in the ~5000 biomedical journals worldwide in the year 2007 alone. The vast majority of these publications include results from hypothesis-driven experimentation in overlapping biomedical research domains. Unfortunately, the sheer volume of information being generated by the biomedical research enterprise has made it virtually impossible for investigators to stay aware of the latest findings in their domain of interest, let alone to (...) be able to assimilate and mine data from related investigations for purposes of meta-analysis. While computers have the potential for assisting investigators in the extraction, management and analysis of these data, information contained in the traditional journal publication is still largely unstructured, free-text descriptions of study design, experimental application and results interpretation, making it difficult for computers to gain access to the content of what is being conveyed without significant manual intervention. In order to circumvent these roadblocks and make the most of the output from the biomedical research enterprise, a variety of related standards in knowledge representation are being developed, proposed and adopted in the biomedical community. In this chapter, we will explore the current status of efforts to develop minimum information standards for the representation of a biomedical experiment, ontologies composed of shared vocabularies assembled into subsumption hierarchical structures, and extensible relational data models that link the information components together in a machine-readable and human-useable framework for data mining purposes. (shrink)

Concept representation is still an open problem in the field of ontology engineering and, more generally, of knowledge representation. In particular, the issue of representing “non classical” concepts, i.e. concepts that cannot be defined in terms of necessary and sufficient conditions, remains unresolved. In this paper we review empirical evidence from cognitive psychology, according to which concept representation is not a unitary phenomenon. On this basis, we sketch some proposals for concept representation, taking into account (...) suggestions from psychological research. In particular, it seems that human beings employ both prototype-based and exemplar-based representations in order to represent non classical concepts. We suggest that a similar, hybrid prototype-exemplar based approach could also prove useful in the field of knowledge representation technology. Finally, we propose conceptual spaces as a suitable framework for developing some aspects of this proposal. (shrink)

We discuss the role of perceptron (or threshold) connectives in the context of Description Logic, and in particular their possible use as a bridge between statistical learning of models from data and logical reasoning over knowledge bases. We prove that such connectives can be added to the language of most forms of Description Logic without increasing the complexity of the corresponding inference problem. We show, with a practical example over the Gene Ontology, how even simple instances of perceptron connectives (...) are expressive enough to represent learned, complex concepts derived from real use cases. This opens up the possibility to import concepts learnt from data into existing ontologies. (shrink)

The interpretation of the abstraction process and the use of various abstractions are consistent with the trends associated with the naturalistic turn in modern cognitive and neural studies. Logic of dealing with abstractions presupposes not only acts of digress from the insignificant details of the object, but also the replenishment of the image due to idealization, endowing the object with properties that are absent from it. Thus, abstraction expresses not only the activity of the subject but the fact of “locking” (...) this activity on a certain kind of ontology as well. The latter, in the spirit of I. Kant’s apriorism, is a function of epistemological attitudes and the nature of the subject's activity. Therefore, in the context of modern neuroscience, we can mean the transcendentalism of activity type. An effective tool for comprehension of abstractions making and development is a metaphor, which, on the one hand, allows submerge the object of analysis into a more or less familiar context, and on the other hand, it may produce new abstractions. Naturalistic tendencies manifested in the fact that empirically established abstractions activate certain neural brain networks, and abstract and concrete concepts are "processed" by various parts of the brain. If we keep in mind the presence of different levels abstractions then not only neural networks but even individual neurons (called “conceptual”) can be excited. The excitation of neural networks is associated with understanding the meaning of some concepts, but at the same time, the activity of these networks presupposes the "dissection" of reality due to a certain angle, determined in the general case by goals, attitudes and concrete practices of the subject. (shrink)

In this inaugural lecture I offer, against the background of a discussion of knowledge representation and its tools, an overview of my research in the philosophy of science. I defend a relational model-theoretic realism as being the appropriate meta-stance most congruent with the model-theoretic view of science as a form of human engagement with the world. Making use of logics with preferential semantics within a model-theoretic paradigm, I give an account of science as process and product. I demonstrate (...) the power of the full-blown employment of this paradigm in the philosophy of science by discussing the main applications of model-theoretic realism to traditional problems in the philosophy of science. I discuss my views of the nature of logic and of its role in the philosophy of science today. I also specifically offer a brief discussion on the future of cognitive philosophy in South Africa. My conclusion is a general look at the nature of philosophical inquiry and its significance for philosophers today. South African Journal of Philosophy Vol. 25 (4) 2006: pp. 275-289. (shrink)

I argue for three points: First, evidence of the primacy of knowledge representation is not evidence of primacy of knowledge. Second, knowledge-oriented mindreading research should also focus on misrepresentations and biased representations of knowledge. Third, knowledge-oriented mindreading research must confront the problem of the gold standard that arises when disagreement about knowledge complicates the interpretation of empirical findings.

The idea that pictorial art can have cognitive value, that it can enhance our understanding of the world and of our own selves, has had many advocates in art theory and philosophical aesthetics alike. It has also been argued, however, that the power of pictorial representation to convey or enhance knowledge, in particular knowledge with moral content, is not generalized across the medium.

(*This paper was awarded the Elisabeth and Werner Leinfellner Award 2017 for outstanding contributions.) -/- This paper provides an explanation of the skeptical puzzle. I argue that we can take two distinct points of view towards representations, mental representations like perceptual experiences and artificial representations like symbols. When focusing on what the representation represents we take an attached point of view. When focusing on the representational character of the representation we take a detached point view. From an attached (...) point of view, we have the intuition that we can know that p simply by using the representation and without having prior knowledge about the reliability of the source that delivers the representation. When taking a detached point of view, we tend to think that we must have this kind of prior knowledge. These two conflicting intuitions about knowledge and representations provide the basis for our intuition of immediate perceptual knowledge on the one hand and for the skeptical intuition of underdetermination on the other hand. (shrink)

Knowledge representation (KR) is actually more than representation: It involves also inference, namely inference of “new” knowledge, i.e. new facts. Logic programming is a suitable KR medium, but more often than not discussions on this programming paradigm focus on aspects other than KR. In this paper, I elaborate on the general theory of logic programming and give the essentials of two of its main implementations, to wit, Prolog and Datalog, from the viewpoint of deductive computing over (...) knowledge bases, which includes deductive programming. (shrink)

In this paper we identify and characterize an analysis of two problematic aspects affecting the representational level of cognitive architectures (CAs), namely: the limited size and the homogeneous typology of the encoded and processed knowledge. We argue that such aspects may constitute not only a technological problem that, in our opinion, should be addressed in order to build arti cial agents able to exhibit intelligent behaviours in general scenarios, but also an epistemological one, since they limit the plausibility of (...) the comparison of the CAs' knowledge representation and processing mechanisms with those executed by humans in their everyday activities. In the fi nal part of the paper further directions of research will be explored, trying to address current limitations and future challenges. (shrink)

A speaker's use of a declarative sentence in a context has two effects: it expresses a proposition and represents the speaker as knowing that proposition. This essay is about how to explain the second effect. The standard explanation is act-based. A speaker is represented as knowing because their use of the declarative in a context tokens the act-type of assertion and assertions represent knowledge in what's asserted. I propose a semantic explanation on which declaratives covertly host a "know"-parenthetical. A (...) speaker is thereby represented as knowing the proposition expressed because that is the semantic contribution of the parenthetical. I call this view parentheticalism and defend that it better explains knowledge representation than alternatives. As a consequence of outperforming assertoric explanations, parentheticalism opens the door to eliminating the act-type of assertion from linguistic theorizing. (shrink)

Research on the capacity to understand others' minds has tended to focus on representations ofbeliefs,which are widely taken to be among the most central and basic theory of mind representations. Representations ofknowledge, by contrast, have received comparatively little attention and have often been understood as depending on prior representations of belief. After all, how could one represent someone as knowing something if one does not even represent them as believing it? Drawing on a wide range of methods across cognitive science, (...) we ask whether belief or knowledge is the more basic kind of representation. The evidence indicates that nonhuman primates attribute knowledge but not belief, that knowledge representations arise earlier in human development than belief representations, that the capacity to represent knowledge may remain intact in patient populations even when belief representation is disrupted, that knowledge (but not belief) attributions are likely automatic, and that explicit knowledge attributions are made more quickly than equivalent belief attributions. Critically, the theory of mind representations uncovered by these various methods exhibits a set of signature features clearly indicative of knowledge: they are not modality-specific, they are factive, they are not just true belief, and they allow for representations of egocentric ignorance. We argue that these signature features elucidate the primary function of knowledge representation: facilitating learning from others about the external world. This suggests a new way of understanding theory of mind – one that is focused on understanding others' minds in relation to the actual world, rather than independent from it. (shrink)

We present a computational analysis of de re, de dicto, and de se belief and knowledge reports. Our analysis solves a problem first observed by Hector-Neri Castañeda, namely, that the simple rule -/- `(A knows that P) implies P' -/- apparently does not hold if P contains a quasi-indexical. We present a single rule, in the context of a knowledge-representation and reasoning system, that holds for all P, including those containing quasi-indexicals. In so doing, we explore the (...) difference between reasoning in a public communication language and in a knowledge-representation language, we demonstrate the importance of representing proper names explicitly, and we provide support for the necessity of considering sentences in the context of extended discourse (for example, written narrative) in order to fully capture certain features of their semantics. (shrink)

In this article we present an advanced version of Dual-PECCS, a cognitively-inspired knowledge representation and reasoning system aimed at extending the capabilities of artificial systems in conceptual categorization tasks. It combines different sorts of common-sense categorization (prototypical and exemplars-based categorization) with standard monotonic categorization procedures. These different types of inferential procedures are reconciled according to the tenets coming from the dual process theory of reasoning. On the other hand, from a representational perspective, the system relies on the hypothesis (...) of conceptual structures represented as heterogeneous proxytypes. Dual-PECCS has been experimentally assessed in a task of conceptual categorization where a target concept illustrated by a simple common-sense linguistic description had to be identified by resorting to a mix of categorization strategies, and its output has been compared to human responses. The obtained results suggest that our approach can be beneficial to improve the representational and reasoning conceptual capabilities of standard cognitive artificial systems, and –in addition– that it may be plausibly applied to different general computational models of cognition. The current version of the system, in fact, extends our previous work, in that Dual-PECCS is now integrated and tested into two cognitive architectures, ACT-R and CLARION, implementing different assumptions on the underlying invariant structures governing human cognition. Such integration allowed us to extend our previous evaluation. (shrink)

We introduce and discuss a knowledge-driven distillation approach to explaining black-box models by means of two kinds of interpretable models. The first is perceptron (or threshold) connectives, which enrich knowledge representation languages such as Description Logics with linear operators that serve as a bridge between statistical learning and logical reasoning. The second is Trepan Reloaded, an ap- proach that builds post-hoc explanations of black-box classifiers in the form of decision trees enhanced by domain knowledge. Our aim (...) is, firstly, to target a model-agnostic distillation approach exemplified with these two frameworks, secondly, to study how these two frameworks interact on a theoretical level, and, thirdly, to investigate use-cases in ML and AI in a comparative manner. Specifically, we envision that user-studies will help determine human understandability of explanations generated using these two frameworks. (shrink)

We describe and try to motivate our project to build systems using both a knowledge based and a neural network approach. These two approaches are used at different stages in the solution of a problem, instead of using knowledge bases exclusively on some problems, and neural nets exclusively on others. The knowledge base (KB) is defined first in a declarative, symbolic language that is easy to use. It is then compiled into an efficient neural network (NN) (...) class='Hi'>representation, run, and the results from run time and (eventually) from learning are decompiled to a symbolic description of the knowledge contained in the network. After inspecting this recovered knowledge, a designer would be able to modify the KB and go through the whole cycle of compiling, running, and decompiling again. The central question with which this project is concerned is, therefore, How do we go from a KB to an NN, and back again? We are investigating this question by building tools consisting of a repertoire of language/translation/network types, and trying them on problems in a variety of domains. (shrink)

Epistemologists have long believed that epistemic luck undermines propositional knowledge. Action theorists have long believed that agentive luck undermines intentional action. But is there a relationship between agentive luck and epistemic luck? While agentive luck and epistemic luck have been widely thought to be independent phenomena, we argue that agentive luck has an epistemic dimension. We present several thought experiments where epistemic luck seems to undermine both knowledge-how and intentional action and we report experimental results that corroborate these (...) judgments. These findings have implications for the role of knowledge in a theory of intentional action and for debates about the nature of knowledge-how and the significance of knowledge representation in folk psychology. (shrink)

Approximation involves representing things in ways that might be close to the truth but are nevertheless false. Given the widespread reliance on approximations in science and everyday life, here we ask whether it is conceptually possible for false approximations to qualify as knowledge. According to the factivity account, it is impossible to know false approximations, because knowledge requires truth. According to the representational adequacy account, it is possible to know false approximations, if they are close enough to the (...) truth for present purposes. In this paper, we adopt an experimental methodology to begin testing these two theories. When an agent provides a false and practically inadequate answer, both theories predict that people will deny knowledge. But the theories disagree about an agent who provides a false but practically adequate answer: the factivity hypothesis again predicts knowledge denial, whereas the representational adequacy hypothesis predicts knowledge attribution. Across two experiments, our principal finding was that people tended to attribute knowledge for false but practically adequate answers, which supports the representational adequacy account. We propose an interpretation of existing findings that preserves a conceptual link between knowledge and truth. According to this proposal, truth is not necessary for knowledge, but it is a feature of prototypical knowledge. (shrink)

Human knowledge is typically described as a linear progression from observation to theory. Historical evidence, cognitive science, and structural analysis suggest that this view is incomplete. Knowledge emerges through a recursive process that begins in symbolic form—fiction, metaphor, narrative, myth—long before formalization becomes possible. This paper proposes Symbolic Pre-Formalism: a rare cognitive mechanism through which individuals express early, partial structures of future theories in symbolic media. These symbolic forms act as compressed prototypes, enabling contradiction-tolerant representation before logical (...) or mathematical articulation is available. -/- The theory integrates classical epistemological models (Plato, Aristotle, Kant, Peirce, Popper, Kuhn, Polanyi, Lakatos) with contemporary insights from cognitive architecture, compression-based reasoning, developmental cognition, and recursive self-reference. Case studies from Goethe, Jung, Borges, Lem, Hesse, and others illustrate the symbolic-to-formal transition across history. The paper then synthesizes these insights into a unified model of six knowledge-emergence modes: sensory compression, symbolic pre-formalism, drift-correction learning, recursive self-reference, social distribution, and deterministic formalization. -/- The resulting framework reframes creativity, scientific discovery, and cognitive development as deterministic, coherence-seeking processes rather than stochastic events. Symbolic forms hold and preserve early structure; formal systems complete it. This unified model clarifies how insight arises, how symbolic narratives anticipate future conceptual frameworks, and why the symbolic → formal recursion is a universal yet uncommon mode of human cognition. (shrink)

It is widely accepted in epistemology that knowledge is factive, meaning that only truths can be known. We argue that this theory creates a skeptical challenge: because many of our beliefs are only approximately true, and therefore false, they do not count as knowledge. We consider several responses to this challenge and propose a new one. We propose easing the truth requirement on knowledge to allow approximately true, practically adequate representations to count as knowledge. In addition (...) to addressing the skeptical challenge, this view also coheres with several previous theoretical proposals in epistemology. (shrink)

This chapter describes Spinoza's obscure “ideas of ideas” doctrine and his claim that “as soon as one knows something, one knows that one knows it, and simultaneously knows that one knows that one knows, and so on, to infinity”. Spinoza holds that the human mind is a representation of the body: the “objectum of the idea constituting the human mind” is the human body. Suppose ideas are essentially self‐reflexive, and that this reflexive awareness, the “idea of the idea,” makes (...) the objectively‐real representational content present to mind. The chapter also suggests that Spinoza holds that adequate ideas come, by nature, with reflective knowledge. Reasoning through the Ethics is one way to come to the adequate ideas that silence doubts about the veridicality of one's adequate ideas of bodies. The “human mind has an adequate cognition of God's eternal and infinite essence”. (shrink)

This paper develops a five-operator framework for transformation-sensitive knowledge, treating knowledge not as detached representation but as participation that stabilizes cognition and aligns models with domain-defining invariants.  The operators—Symbolic Mediation, Compositional Correspondence, Attentional Stabilization, Participatory Alignment, and Reflexive Fixpoint—come with minimal mathematical assumptions and auditable acceptance tests (convergence, invariant alignment, independence-based reflexive checks). A worked toy example and a concise reporting protocol (“Reproducibility Card”) show how to estimate contraction, bound cross-level distortion, and distinguish robust fixpoints from (...) self-sealing coherence. Historically, the scheme is structurally inspired by late-antique and Renaissance Hermetic motifs (logos, correspondence, disciplined practice, participation, unity) without making causal claims. The result is a domain-independent method that makes persistent insight empirically separable from transient impression and is readily instantiated across logic, modeling, and cognition, with stated scope conditions and threats to validity. (shrink)

The extraordinary success of recent Large Language Models (LLMs) on a diverse array of tasks has led to an explosion of scientific and philosophical theorizing aimed at explaining how they do what they do. Unfortunately, disagreement over fundamental theoretical issues has led to stalemate, with entrenched camps of LLM optimists and pessimists often committed to very different views of how these systems work. Overcoming stalemate requires agreement on fundamental questions, and the goal of this paper is to address one such (...) question, namely: is LLM behavior driven partly by representation-based information processing of the sort implicated in biological cognition, or is it driven entirely by processes of memorization and stochastic table look-up? This is a question about what kind of algorithm LLMs implement, and the answer carries serious implications for higher level questions about whether these systems have beliefs, intentions, concepts, knowledge, and understanding. I argue that LLM behavior is partially driven by representation-based information processing, and then I describe and defend a series of practical techniques for investigating these representations and developing explanations on their basis. The resulting account provides a groundwork for future theorizing about language models and their successors. (shrink)

The standard representation theorem for expected utility theory tells us that if a subject’s preferences conform to certain axioms, then she can be represented as maximising her expected utility given a particular set of credences and utilities—and, moreover, that having those credences and utilities is the only way that she could be maximising her expected utility. However, the kinds of agents these theorems seem apt to tell us anything about are highly idealised, being always probabilistically coherent with infinitely precise (...) degrees of belief and full knowledge of all a priori truths. Ordinary subjects do not look very rational when compared to the kinds of agents usually talked about in decision theory. In this paper, I will develop an expected utility representation theorem aimed at the representation of those who are neither probabilistically coherent, logically omniscient, nor expected utility maximisers across the board—that is, agents who are frequently irrational. The agents in question may be deductively fallible, have incoherent credences, limited representational capacities, and fail to maximise expected utility for all but a limited class of gambles. (shrink)

This paper develops Wu’s Three Principles of Complexity Epistemology, a framework designed to explain why human knowledge, scientific theories, and artificial intelligence systems systematically fail to achieve monotonic growth, axiomatic closure, or global consistency. I begin by arguing that all epistemic activity involves mapping an open, continuous, and nonlinear domain D into a discrete representational system R. This structural mismatch forces representational compression and generates unavoidable information loss. From this, I derive three epistemic principles. The Anti-Monotonicity Principle (AMP) states (...) that epistemic progress cannot proceed through linear accumulation: new distinctions in D often invalidate prior commitments in R, yielding intrinsically revisionary theory change. The Anti-Closure Principle (ACP) generalizes Gödelian incompleteness to all open domains, showing that no recursively enumerable system can exhaustively encode an indefinitely extensible target. The Tension Principle (TP) explains why representational systems routinely exhibit contradictions, dualities, or structural incompatibilities: compression forces distinct domain-level structures into overlapping symbolic forms whose inferential roles may conflict. I demonstrate the explanatory power of this framework through four case studies: Gödel’s incompleteness theorems, the Einsteinian restructuring of spacetime, the recurrent failure of utopian social designs, and the systematic limitations of large language models. I further address potential objections concerning self-reference, relativism, over-formalization, ontological commitment, and instrumentalism. The resulting picture is not skeptical but post-rationalist: rational inquiry remains possible and objective, yet all representational systems are structurally partial, dynamically revisable, and capable of sustaining localized tension. The paper concludes by suggesting how this framework reframes scientific methodology, AI research, and epistemic normativity under conditions of complexity. (shrink)

This article investigates the epistemological nature of the act of believing, aiming to clarify to what extent belief can be considered a legitimate instance of knowledge or a mental state with epistemic value. The analysis begins with the conception of belief as a representational psychological disposition associated with a predicative proposition, whose validity and truth are presupposed by the subject. It is argued that, for a belief to possess full epistemic status, it must be properly justified and true, resulting (...) from a cognitive process oriented toward reality. The implications of Gettier-style cases and Alvin Plantinga's approach to warrant are examined, confronting internalist and externalist perspectives on justification. It is concluded that believing, although psychologically necessary, is epistemically valid only when it is open to the rational and ontological scrutiny of truth, and may, under certain conditions, coincide with knowledge. Belief is thus conceived as a mental state that participates in the economy of knowledge, but whose epistemic value is conditional, graded, and dependent on its grounding in reality. (shrink)

Amodal completion is the representation of occluded parts of perceived objects. We argue for the following three claims: First, at least some amodal completion-involved experiences can ground knowledge about the occluded portions of perceived objects. Second, at least some instances of amodal completion-grounded knowledge are not sensitive, that is, it is not the case that in the nearest worlds in which the relevant claim is false, that claim is not believed true. Third, at least some instances of (...) amodal completion-grounded knowledge are not safe, that is, it is not the case that in all or nearly all near worlds where the relevant claim is believed true, that claim is in fact true. Thus, certain instances of amodal completion-grounded knowledge refute both the view that knowledge is necessarily sensitive and the view that knowledge is necessarily safe. (shrink)

Review of Joseph Y. Halpern (ed.), Theoretical Aspects of Reasoning About Knowledge: Proceedings of the 1986 Conference (Los Altos, CA: Morgan Kaufmann, 1986),.

The engineering knowledge research program is part of the larger effort to articulate a philosophy of engineering and an engineering worldview. Engineering knowledge requires a more comprehensive conceptual framework than scientific knowledge. Engineering is not ‘merely’ applied science. Kuhn and Popper established the limits of scientific knowledge. In parallel, the embrace of complementarity and uncertainty in the new physics undermined the scientific concept of observer-independent knowledge. The paradigm shift from the scientific framework to the broader (...) participant engineering framework entails a problem shift. The detached scientific spectator seeks the ‘facts’ of ‘objective’ reality – out there. The participant, embodied in reality, seeks ‘methods’, about how to work in the world. The engineering knowledge research program is recursively enabling. Advances in engineering knowledge are involved in the unfolding of the nature of reality. Newly understood, quantum uncertainty entails that the participant is a natural inquirer. ‘Practical reason’ is concerned with ‘how we should live’– the defining question of morality. The engineering knowledge research program is selective seeking ‘important truths’, ‘important knowledge’, ‘important methods’ that manifest value, and serve the engineering agenda of ‘the construction of the good.’ The importance of engineering knowledge research program is clear in the new STEM curriculum where educators have been challenged to rethink the relation between science and engineering. A 2015 higher education initiative to integrate engineering colleges into liberal arts and sciences colleges has stalled due to the confusion and conflict between the engineering and scientific representations of knowledge. (shrink)

A. Newell and H. A. Simon were two of the most influential scientists in the emerging field of artificial intelligence (AI) in the late 1950s through to the early 1990s. This paper reviews their crucial contribution to this field, namely to symbolic AI. This contribution was constituted mostly by their quest for the implementation of general intelligence and (commonsense) knowledge in artificial thinking or reasoning artifacts, a project they shared with many other scientists but that in their case was (...) theoretically based on the idiosyncratic notions of symbol systems and the representational abilities they give rise to, in particular with respect to knowledge. While focusing on the period 1956-1982, this review cites both earlier and later literature and it attempts to make visible their potential relevance to today's greatest unifying AI challenge, to wit, the design of wholly autonomous artificial agents (a.k.a. robots) that are not only rational and ethical, but also self-conscious. (shrink)

Several philosophers have proposed Knowledge-Based Decision Theories (KDTs)—theories that require agents to maximize expected utility as yielded by utility and probability functions that depend on the agent’s knowledge. Proponents of KDTs argue that such theories are motivated by Knowledge-Reasons norms that require agents to act only on reasons that they know. However, no formal derivation of KDTs from Knowledge-Reasons norms has been suggested, and it is not clear how such norms justify the particular ways in which (...) KDTs relate knowledge and rational action. In this paper, I suggest a new axiomatic method for justifying KDTs and providing them with stronger normative foundations. I argue that such theories may be derived from constraints on the relation between knowledge and preference, and that these constraints may be evaluated relative to intuitions regarding practical reasoning. To demonstrate this, I offer a representation theorem for a KDT proposed by Hawthorne and Stanley (2008) and briefly evaluate it through its underlying axioms. (shrink)

[Significantly updated in Chapter 7 of Awareness and the Substructure of Knowledge] Many have found it plausible that knowledge is a constitutively normative state, i.e. a state that is grounded in the possession of reasons. Many have also found it plausible that certain cases of proprioceptive knowledge, memorial knowledge, and self-evident knowledge are cases of knowledge that are not grounded in the possession of reasons. I refer to these as cases of basic knowledge. (...) The existence of basic knowledge forms a primary objection to the idea that knowledge is a constitutively normative state. In what follows I offer a way through the apparent dilemma of having to choose between either basic knowledge or the normativity of knowledge. The solution involves homing in on a state of awareness (≈non-accidental true representation) that is distinct from knowledge and which in turn grounds the normativity of knowledge in a way that is fully consistent with the existence of basic knowledge. An upshot of this is that externalist theories of knowledge turn out to be fully compatible with the thesis that knowledgeable beliefs are always beliefs that are justified by the reasons one possesses. (shrink)

Much recent cognitive neuroscientific work on body knowledge is representationalist: “body schema” and “body images”, for example, are cerebral representations of the body (de Vignemont 2009). A framework assumption is that representation of the body plays an important role in cognition. The question is whether this representationalist assumption is compatible with the variety of broadly situated or embodied approaches recently popular in the cognitive neurosciences: approaches in which cognition is taken to have a ‘direct’ relation to the body (...) and to the environment. A “direct” relation is one where the boundaries between the body and the head, or between the environment and the animal are not theoretically important in the understanding of cognition. These boundaries do not play a theoretically privileged role in cognitive explanations of behavior. But representationalism appears to put a representational veil between the locus of cognition and that which is represented, making cognitive relations to the body and to the environment be indirect, with a high associated computational load. For this reason, direct approaches have tried to minimize the use of internal representations (Suchman 1987; Barwise 1987; Agre and Chapman 1987; Brooks 1992; Thelen and Smith 1994; van Gelder 1995; Port and van Gelder 1995; Clark 1997, 1999; Rupert 2009, p. 180). Does a cognitive neuroscience committed to direct relations rule out a representationalist approach to body knowledge? Or is direct representationalism possible? (shrink)

I discuss the difference and the connections between Kant’s notions of cognition (Erkenntnis) and knowledge (Wissen). Unlike knowledge, cognition is a representational state which need not have the propositional structure of a judgments. Even cognitions that have such a structure need not coincide with knowledge, because they might rather have the doxastic status of opinion or faith, or they might be false (whereas knowledge is a certain recognition of truth). I argue that while Kant distinguishes between (...) many different species of cognition, he uses the term ‘knowledge’ univocally across his theoretical and practical philosophy. All knowledge is based upon some kind of cognition that provides the sufficient objective, certain ground for a knowing assent to a proposition. However, not all knowledge must be based upon the demanding type of cognition that is central to Kant’s overall argument in the first Critique, which requires a synthesis of concepts and sensible intuition. We have a lot of determinate knowledge, including knowledge that we possess transcendental freedom of will, that is based upon purely conceptual practical cognition. We can also have some indeterminate philosophical knowledge (e.g. knowledge that there are non-sensible, noumenal beings) that is based upon purely conceptual theoretical cognition. (shrink)

Although other animals can make simple tools, the expanded and complex material culture of humans is unprecedented in the animal kingdom. Tool making is a slow and late-developing ability in humans, and preschool children find making tools to solve problems very challenging. This difficulty in tool making might be related to the lack of familiarity with the tools and may be overcome by children's long term perceptual-motor knowledge. Thus, in this study, the effect of tool familiarity on tool making (...) was investigated with a task in which 5-to-6-year-old children (n = 75) were asked to remove a small bucket from a vertical tube. The results show that children are better at tool making if the tool and its relation to the task are familiar to them (e.g., soda straw). Moreover, we also replicated the finding that hierarchical complexity and tool making were significantly related. Results are discussed in light of the ideomotor approach. (shrink)

This work aims to contribute to the explanation of cognitive capacities that are essential to human intelligence: commonsense and situation holism. The attempt to explain them within the field of cognitive sciences raises a foundational challenge. How can human cognition distinguish what’s relevant and what’s not in an open-ended set of contexts? The challenge is characterized by a circularity. Potential solutions end up relying on the very capacity that they should be explaining, i.e. the sensitivity to what’s contextually relevant. The (...) argument here developed is an attempt to get out of this circularity. It is grounded on a bi-dimensional analysis of the problem. The first one is representational and is about how cognitive knowledge on what’s relevant can be stored. This dimension is related to the frame problem from artificial intelligence. I claim that this representational aspect of the issue can be neutralized by abandoning language-like representational schemes and sticking to a plurality of structural ones. They allow representational states to play a clear and non-trivial explanatory role while enabling a kind of representational productivity that is not bothered by the frame problem. The second dimension is the inferential one. It encompasses our relevance-sensitive inferential productivity, i.e. our capacity to efficiently infer what’s relevant in indefinitely many contexts, including ones we’ve never faced before. I argue that such relevance sensitivity cannot be neither innate, for innate mechanisms result in the aforementioned circularity, nor learned through innate mechanisms, for learning processes themselves rely on relevance sensitivity. The suggested alternative is that relevance sensitivity is a cognitive gadget. A kind of mental mechanism that is culturally, rather than genetically, inherited. This possibility is rendered plausible within Cecilia Heyes’ “cognitive gadgets” framework. She shows how developmental tweaks and biases culturally established can participate in the constitution of cognitive mechanisms. Thus, cultural elements can participate in human cognition not only as information sources, but also in the form of tweaks that bias the processing of that information towards what’s culturally established as contextually relevant. (shrink)

During the last decades, many cognitive architectures (CAs) have been realized adopting different assumptions about the organization and the representation of their knowledge level. Some of them (e.g. SOAR [35]) adopt a classical symbolic approach, some (e.g. LEABRA[ 48]) are based on a purely connectionist model, while others (e.g. CLARION [59]) adopt a hybrid approach combining connectionist and symbolic representational levels. Additionally, some attempts (e.g. biSOAR) trying to extend the representational capacities of CAs by integrating diagrammatical representations and (...) reasoning are also available [34]. In this paper we propose a reflection on the role that Conceptual Spaces, a framework developed by Peter G¨ardenfors [24] more than fifteen years ago, can play in the current development of the Knowledge Level in Cognitive Systems and Architectures. In particular, we claim that Conceptual Spaces offer a lingua franca that allows to unify and generalize many aspects of the symbolic, sub-symbolic and diagrammatic approaches (by overcoming some of their typical problems) and to integrate them on a common ground. In doing so we extend and detail some of the arguments explored by G¨ardenfors [23] for defending the need of a conceptual, intermediate, representation level between the symbolic and the sub-symbolic one. In particular we focus on the advantages offered by Conceptual Spaces (w.r.t. symbolic and sub-symbolic approaches) in dealing with the problem of compositionality of representations based on typicality traits. Additionally, we argue that Conceptual Spaces could offer a unifying framework for interpreting many kinds of diagrammatic and analogical representations. As a consequence, their adoption could also favor the integration of diagrammatical representation and reasoning in CAs. (shrink)

This essay presents a philosophical and computational theory of the representation of de re, de dicto, nested, and quasi-indexical belief reports expressed in natural language. The propositional Semantic Network Processing System (SNePS) is used for representing and reasoning about these reports. In particular, quasi-indicators (indexical expressions occurring in intentional contexts and representing uses of indicators by another speaker) pose problems for natural-language representation and reasoning systems, because--unlike pure indicators--they cannot be replaced by coreferential NPs without changing the meaning (...) of the embedding sentence. Therefore, the referent of the quasi-indicator must be represented in such a way that no invalid coreferential claims are entailed. The importance of quasi-indicators is discussed, and it is shown that all four of the above categories of belief reports can be handled by a single representational technique using belief spaces containing intensional entities. Inference rules and belief-revision techniques for the system are also examined. (shrink)

Many have urged that the biggest obstacles to a physicalistic understanding of consciousness are the problems raised in connection with the subjectivity of consciousness. These problems are most acutely expressed in consideration of the knowledge argument against physicalism. I develop a novel account of the subjectivity of consciousness by explicating the ways in which mental representations may be perspectival. Crucial features of my account involve analogies between the representations involved in sensory experience and the ways in which pictorial representations (...) exhibit perspectives or points of view. I argue that the resultant account of subjectivity provides a basis for the strongest response physicalists can give to the knowledge argument. (shrink)

Since the nineteen-seventies, knowledge engineering as a discipline has struggled with the implementation problem: the difficulty of translating expert knowledge expressed in natural language into a formal knowledge representation to be adopted by organizations and communities for use in automated decision making. This knowledge acquisition bottleneck remains a fundamental barrier. We argue that large language models (LLMs) provide a means to address the implementation problem, by allowing knowledge expressed in natural language to be used (...) directly in knowledge engineering tasks, rather than having to be first formalized into a knowledge representation language. We show how classifiers-as-intensions, based on the prompting of LLMs using natural language intensional definitions of concepts and relations, can provide support for the knowledge engineering task of classification. We show that by having classifiers-as-intensions provide rationales for their classifications, we can distinguish factual errors from disagreements about the meaning of concepts and relations, yielding actionable guidance for knowledge graph refinement. One objection to this approach is that LLMs exhibit hallucination in their output, bringing into question their factuality. To address this objection, we show that LLMs are capable of accurately detecting hallucination in language model output, and that bilateral factuality evaluation provides insight into the degree and scope of inconsistency and incompleteness in an LLM's parametric knowledge. We show how bilateral factuality evaluation can then be used in the formal semantics of a paraconsistent logic to allow sound and complete neurosymbolic reasoning using such knowledge. We conclude by arguing that the implementation problem in knowledge engineering is rooted in its adherence to representationalism, and that our findings suggest that inferentialism and social externalism provide a way to reconceptualize the practice of knowledge engineering and dissolve the implementation problem, not by making LLMs reason logically, but by using logics that allow reasoning with LLMs. (shrink)

The modelling of concepts, besides involving disciplines like philosophy of mind and psychology, is a fundamental and lively research problem in several artificial intelligence (AI) areas, such as knowledge representation, machine learning, and natural language processing. In this scenario, the most prominent proposed solutions adopt different (often incompatible) assumptions about the nature of such a notion. Each of these solutions has been developed to capture some specific features of concepts and support some specific (artificial) cognitive operations. This paper (...) critically reviews the most notable computational approaches to the representation of concepts. The main goals are (i) to provide a shared terminology for the desiderata of concepts and their computational representation; (ii) to classify and assess the heterogeneous computational approaches according to the provided terminology; (iii) to provide a reader who may not be very familiar with theories of concepts with an introduction to major themes in this research and with pointers to different research projects, and (iv) to offer philosophers, and potentially AI practitioners, a well-informed guide for selecting among various (and possibly competing) computational representations of concepts. (shrink)

A 3rd person Knowledge Level analysis of cognitive architectures -/- Abstract I provide a knowledge level analysis of the main representational and reasoning problems affecting the cognitive architectures for what concerns this issue. In providing this analysis I will show, by considering some of the main cognitive architectures currently available (e.g. SOAR, ACT-R, CLARION), how one of the main problems of such architectures is represented by the fact that their knowledge representation and processing mechanisms are not (...) sufficiently constrained with “structural insights” (Lieto 2021) coming from cognitive science for dealing with commonsense knowledge and reasoning (Lebiere, Oltramari, 2018). As a possible way out to such knowledge processing issues, I present the main assumptions that have led to the development of the Dual PECCS categorization system (Lieto, Radicioni, Rho 2017) and discuss some of the lessons learned and their possible implications in the design of the knowledge modules and knowledge-processing mechanisms of integrated cognitive architectures. (shrink)

Formal ontologies are nowadays widely considered a standard tool for knowledge representation and reasoning in the Semantic Web. In this context, they are expected to play an important role in helping automated processes to access information. Namely: they are expected to provide a formal structure able to explicate the relationships between different concepts/terms, thus allowing intelligent agents to interpret, correctly, the semantics of the web resources improving the performances of the search technologies. Here we take into account a (...) problem regarding Knowledge Representation in general, and ontology based representations in particular; namely: the fact that knowledge modeling seems to be constrained between conflicting requirements, such as compositionality, on the one hand and the need to represent prototypical information on the other. In particular, most common sense concepts seem not to be captured by the stringent semantics expressed by such formalisms as, for example, Description Logics (which are the formalisms on which the ontology languages have been built). The aim of this work is to analyse this problem, suggesting a possible solution suitable for formal ontologies and semantic web representations. The questions guiding this research, in fact, have been: is it possible to provide a formal representational framework which, for the same concept, combines both the classical modelling view (accounting for compositional information) and defeasible, prototypical knowledge ? Is it possible to propose a modelling architecture able to provide different type of reasoning (e.g. classical deductive reasoning for the compositional component and a non monotonic reasoning for the prototypical one)? We suggest a possible answer to these questions proposing a modelling framework able to represent, within the semantic web languages, a multilevel representation of conceptual information, integrating both classical and non classical (typicality based) information. Within this framework we hypothesise, at least in principle, the coexistence of multiple reasoning processes involving the different levels of representation. (shrink)

I examine two explanatory models of choking: the representationalist model and the anti-representationalist model. The representationalist model is based largely on Anderson's ACT model of procedural knowledge and is developed by Masters, Beilock and Carr. The antirepresentationalist model is based on dynamical models of cognition and embodied action and is developed by Dreyfus who employs an antirepresentational view of know-how. I identify the models' similarities and differences. I then suggest that Dreyfus is wrong to believe representational activity requires reflection (...) and attention. I also argue that the representationalist model of choking is preferable, since some embodied actions require appeals to representations, something not available to Dreyfus's anti-representational model. (shrink)

In this paper a possible general framework for the representation of concepts in cognitive artificial systems and cognitive architectures is proposed. The framework is inspired by the so called proxytype theory of concepts and combines it with the heterogeneity approach to concept representations, according to which concepts do not constitute a unitary phenomenon. The contribution of the paper is twofold: on one hand, it aims at providing a novel theoretical hypothesis for the debate about concepts in cognitive sciences by (...) providing unexplored connections between different theories; on the other hand it is aimed at sketching a computational characterization of the problem of concept representation in cognitively inspired artificial systems and in cognitive architectures. (shrink)

Much recent work elucidates different types of representational format, and ways that aspects of perception and cognition may be formatted. Our paper targets an underexplored topic: the format of motor representations, the psychological states that serve as the primary causal link between an agent’s immediate intention to act and their subsequent behaviour. In section 2, we situate motor representations within the context of processes of motor planning and motor control. In section 3, we discuss key differences between symbolic and analog (...) representational format types. In section 4, we argue that there is evidence indicating that motor representations show a core marker of analog formatting. In section 5, we argue that motor representations show core markers of symbolic formatting. In section 6, we characterize motor representations as an interesting form of hybrid representation. In a slogan, motor representation is a symbolic scheme that utilizes analog representations as cognitive tools. We then highlight several open questions this characterization raises and note how it may impact a range of philosophical debates, such as those pertaining to the relationship between propositional knowledge and skilled action guidance, the interface problem, and the issue of how to draw boundaries between central cognition and the motoric. (shrink)