Theory of Mind and the Evolution of Cognition
Emeritus Professor Peter Gärdenfors
Apr 23, 2021
Theory of mind is not a unitary phenomenon. In the first part of the talk I break it down into the roles of reading the emotions, the attention, the desires, the intentions and the beliefs of others. I present some evidence about experimental studies concerning these components in children and in non-human animals. In the second part I apply this analysis to the evolution of cognition. First, I talk about how various components of theory of mind can be used to understand the evolution of causal cognition. Second, I briefly mention the role of theory of mind in the evolution of teaching. Third, I show how the components can be applied when analyzing the evolution of communication.
Neo-Expressivism and the Matching Principle
Apr 1, 2021
When an agent first-personally self-ascribes—avows—a mental state, it is appropriate to presume the avowal’s truth. One way of explaining such “first-person authority” is in terms of what avowals express and are taken to express. According to neo-expressivism, the relevant expressed feature is the very mental state self-ascribed (Bar-On 2000, 2004). Thus, avowals are first-person authoritative because they are actions that (are taken to) express the very mental states that they semantically represent. Put differently, avowals are first-person authoritative insofar as they (are taken to) match in the mental states they express and their semantic contents. Matthew Parrott (2015) has argued that this ‘matching principle’ is vulnerable to several objections. I respond to these objections on behalf of neo-expressivism. My conclusion is that neo-expressivism is not objectionable in virtue of embracing this matching principle.
Investigating the Interplay between Morphosyntax and Memory for Events: The Case of Past Participles
Feb 25, 2021
The representational product of sentence comprehension is the result of the interplay between episodic and semantic memory and our knowledge of the grammatical devices of our language which guide how we retrieve information from these systems. Past participles, being a part of speech derived from verbs but used in a prenominal position (e.g. words like dried) create an interesting test case to examine the effect of morphosyntax on sentence-level meaning representation: on the one hand, they share eventive attributes with state change verbs and, on the other hand, they serve a modifying function, similarly to adjectives. The present work investigated the representational content of prenominal past participles. We considered two hypotheses. The eventive reading hypothesis predicts that participles activate both the resulting and the initial states entailed by the root verb they are derived from. The stative reading hypothesis predicts activation of only the linguistically denoted state. What distinguishes between these two accounts is activation of the initial state, which we took as a marker of eventivity. Two behavioral sentence-picture verification tasks examined the meaning of participles by testing how internal state representations instantiated during reading map onto the external ones presented at the sentence offset. Additionally, an fMRI sentence-comprehension task examined the meaning of participles during the course of language comprehension, where state representations are evoked by the unfolding language. Across two behavioral experiments, we demonstrated reduced accessibility of the initial state following participles and ruled out the interpretation that the initial state is fully blocked. The fMRI experiment further refined this interpretation by suggesting that there is interference between two incompatible state representations and active suppression of the situationally-irrelevant initial state. All three experiments also replicated the finding that both the initial and end states are accessible after state change verbs. We conclude with a discussion of the role of morphosyntax in controlling the interplay between episodic and semantic memory.
The Communicative Foundations of Propositional Attitude Psychology
Jan 28, 2021
According to a widely held dogma, a developed propositional attitude psychology is a prerequisite of attributing communicative intent, and so a developmental prerequisite of natural language acquisition. This view is difficult to reconcile with developmental evidence, which shows not only that children do not develop propositional attitudes until they are four years old (e.g. Rakoczy 2017), but also that this development is parasitic upon natural language acquisition (de Villiers & de Villers 2000; Lohmann & Tomasello 2003; Low 2010), and that it recruits brain regions that do not exist in infancy (Grosse-Wiesmann et al. 2017). Against the received view, and building on my work on minimally Gricean communication (Moore 2017a), I sketch a developmental trajectory to show how propositional attitude psychology could be both invented and learned through communicative interaction. I finish by considering the conditions in which cultural tools for mental state representation might first have been developed in human history; and the extent to which our early human ancestors might have lacked propositional attitudes. The goal of the paper will not be to show that strong nativism about human mindreading must be false, but that there is no reason to take it for granted in considering the origins of the modern human mind.
Minimal Theory of Mind – A Millikanian Approach
Dec 3, 2020
Minimal theory of mind (ToM) is presented in the theory of mind literature as a middle ground between full-blown ToM and behavior-reading. Minimal ToM seems to be a useful construct for studying and understanding the minds of nonhuman animals and infants. However, providing an account of minimal ToM on which minimal mindreading is less cognitively demanding than full-blown mindreading yet more than just a behavior-reading process is a challenge. In this talk, I will consider the possibility of using Ruth Millikan’s theory of mental representation as a framework within which to provide an account of minimal ToM. First, I will present a recent account of minimal ToM that uses Millikan’s theory to establish minimal ToM as a middle ground between full-blown mindreading and behavior-reading. Then I will critically evaluate this account and argue that it fails to make minimal mindreading less cognitively demanding than full-blown mind-reading in any significant way. Finally, I will present an alternative Millikanian account of minimal ToM on which minimal mindreading is not too cognitively demanding, and offer it as a promising account worthy of further consideration.
Oct 15, 2020
There are several ways in which one can be an expressivist about epistemic modality. A rough and general way of characterizing them is to say that modalized sentences express, in some sense, an attitude of the speaker toward the prejacent of the modal. This core idea can be implemented in several ways, the common core of which is that epistemic modals do not have any (significant) effect on the conversational common ground. These ways are typically tested against a number of data points. In this presentation, I will present a few more data points which show that epistemic modals do have a more far-reaching pragmatic effect than originally recognized. This might constitute evidence to move further away, if not break, from expressivist paradigms.
Proper Function and Ethical Judgment: Towards a Biosemantic Account of Ethical Thought and Discourse
Sept 24, 2020
What, if anything, do ethical judgments represent? What role do ethical judgments and ethical claims play in the lives of social creatures like us? This paper employs Ruth Millikan's biosemantic theory to explore answers to these questions through investigating the proper function of ethical judgments and claims. According to the account I propose, ethical judgments and claims are a species of "Pushmi-Pullyu Representation" (Millikan 1995), having at once directive and descriptive directions of fit with the world. In particular, I argue that ethical claims and judgments simultaneously describe some morally salient features of a situation, and direct fitting action in response to those features, in service of a particular kind of distal social coordination function. This proposal has implications for several core topics in metaethics, including the cognitivism/non-cognitivism debate, motivational internalism, and the origins of moral cognition.
Apr 09, 2019
It is a truth universally acknowledged that imagination is a primitive mental state type, irreducible to other mental state types. This is, at least, “one of four basic claims about imagination that enjoy near universal agreement” (Kind, 2016). I will challenge this orthodox view, arguing that imagination can in fact be reduced to, and explained in terms of, one’s being other kinds of familiar folk psychological mental states. The full case for this account is developed in a book I am now completing. Today I simply aim to clear space for the approach by undermining the most commonly voiced—and seemingly most decisive—reasons for thinking that imagination is irreducible to other mental state types.
Feb 28, 2019
John Mackie famously wrote:
Another way of bringing out this queerness is to ask, about anything that is supposed to have some objective moral quality, how this is linked with its natural features. What is the connection between the natural fact that an action is a piece of deliberate cruelty—say, causing pain just for fun—and the moral fact that it is wrong? … The wrongness must somehow be ‘consequential’ or ‘supervenient’; it is wrong because it is a piece of deliberate cruelty. But just what in the world is signified by this ‘because’?
Mackie's challenge is, of course, always taken to be a challenge for Nonnaturalist Moral Realism. For Expressivists, the question of what in the world is signified by the queer 'because' is misguided. But, it is surprisingly difficult to say, in Expressivist terms, what that queer 'because' does mean.
Feb 15, 2019
A surge of empirical research demonstrating flexible cognition in animals and young infants has raised interest in the possibility of rational decision-making in the absence of language. A venerable position, which I here call “Classical Inferentialism”, holds that nonlinguistic agents are incapable of rational inferences. In the first half of this talk, I argue against this position, defending a model of nonlinguistic inferences that shows how nonlinguistic agents could be practically rational. This model vindicates the Lockean idea that we can intuitively grasp rational connections between thoughts by developing the Davidsonian idea that practical inferences are at bottom categorization judgments. From this perspective, we can see how similarity-based categorization processes widely studied in human and animal psychology might count as practically rational. The solution involves a novel hybrid of internalism and externalism: intuitive inferences are psychologically rational (in the explanatory sense) given the intensional sensitivity of the similarity assessment to the internal structure of the agent’s reasons for acting, but epistemically rational (in the justificatory sense) given an ecological fit between the features matched by that assessment and the structure of the agent’s environment. A key insight is that many animals are sensitive to the strength of their intuitive hunches, and can exert executive control to inhibit responding to some options in favor of others that are more appropriate to their goals.
The second half of the talk explores the neural mechanisms that could implement these forms of sophisticated similarity-assessment and control, and how far they might be scaled up even without language. In mammals, it is argued that the neocortex possesses powerful mechanisms for learning even very highly abstract similarities common amongst members of a category, which grounds the ability of animals to display “insightful” causal or social inferences which appear independent of any specific stimulus associations. The nature of this abstraction ability will be explained by reference to deep learning neural network models, whose architectures have been explicitly designed to model these forms of cortical processing. A potential complication is that some of the most cognitively flexible animals are birds, which lack a neocortex entirely. The talk will end with a discussion of the relevant differences between the neuroanatomy of mammals and birds, appealing to the latest findings in avian neuroscience to speculate how their brains might have arrived at a very spatial organization that achieves functionally similar processing under very different biophysical constraints--providing an interesting case study for traditional philosophical concerns about multiple realizability.
Jill de Villiers
Dec 7, 2018
I have been pursuing ideas about how language helps us think, a position compatible with Chomsky’s position that the major function of language in the species is a cognitive one. Although it is fairly obvious that language helps us scaffold long chains of reasoning – and writing does this even more – is there something more fundamental that language gives us? Is it just words, and therefore symbols, that do this, or does syntax have a role to play? I briefly review evidence from child development that the acquisition of language –any language- may change the way children think, considering work in domains of sortals, number, space, and causation. On the basis of my own empirical work with infants and children, and with adults in situations that interfere with their use of the language faculty, I suggest that syntax may provide us with fundamental new ways of encoding events, and therefore new kinds of concepts.
Nov 6, 2018
One of the primary differences between language production and nonhuman primate vocal production is that language conveys meaning through the hierarchical structuring of elements. A longstanding question concerns the evolutionary origin of the constraints on language sequencing, in particular because nonhuman primate vocal repertoires are largely fixed. In this talk, I present research from our lab and others that investigates sequential motor planning abilities in several primate species, including humans, in effort to explore whether this domain holds promise for understanding the origins of constraints on language production.
Nov 5, 2018
Infant and adult learners possess a remarkable ability to track distributional regularities available in sensory input. Proficiency with statistical learning is important for the early stages of language acquisition and also correlates with second language learning outcomes in adults. However, much of the research in this field has implicitly assumed that the task of the learner is to acquire a single, uniform pattern. Given that real world learning affords far more variability, this simplifying assumption may obscure a significant problem confronting learners: determining how many underlying structures are generating the input. We have found that learners can track statistics in a contextually sensitive manner, and that their bias toward stationarity (i.e., expecting a single underlying structure) can be overcome. Nevertheless, there are both costs and benefits associated with being an efficient learner. In the second part of the talk, I will discuss the growing literature on bilingualism and statistical learning, arguing that experience with multiple languages may shape expectations, but does not alter core learning abilities.
Escape from Fraughtness: How Signaling Emerges as an Aspect of Coordination in a Cooperative Game among Humans
Oct 12, 2018
How do societal communication systems (like human languages) first get started? An interesting approach to this question, called experimental semiotics, involves running experiments in which groups of people are asked to engage in a cooperative task, and then seeing what kind of communication systems emerge. Most experiments in this domain set up a referential situation (e.g., try to get your partner to point to an object shown just to you from among a larger array that she can see) using a predefined communication channel (e.g., playing tunes on a pennywhistle). We note that this type of set-up presupposes that the idea of communication is already understood by community members and that they identify it as a goal. In the present work, we push on the question, “How might such social-cognitive conditions get established in the first place?” We describe an experiment with multiple participants engaging in a coordination task where, initially, participants are simply acting not talking. Because they can observe each other’s actions, they are immediately communicating in a very simple sense. Later, a particularly challenging situation (called fraughtness) arises. Under this condition, we find evidence that a slightly more complex type of action/communication emerges. This turns out to resemble a kind of communication that honeybees do when swarming. We argue that this is the kind of case we should focus on if the goal is to understand the origins of language.
May 4, 2018
In this talk, I present an evolutionary challenge to a widely accepted cognitivist account of communication growing out of the work of H.P. Grice. This evolutionary challenge, as I shall develop it, turns on both comparative data on animal communication systems and on the kinds of selective pressures that would drive a population of agents to be capable of what Grice calls SPEAKER MEANING. In the first half of the talk, I argue that this challenge is serious—affecting Grice’s original 1957 position, as well as subsequent refinements on Grice’s ideas such as the Relevance Theoretic framework of Deidre Wilson and Dan Sperber as well as the pragmatic framework of Robert Stalnaker. In the second half of the talk, I argue that the challenge is nonetheless surmountable. In particular, I develop a perspective on communication that centers on a primitive socio-cognitive relation that I call REPRESENTATIONAL COORDINATION. I argue that this socio-cognitive relation is well suited to understand the origins of communication in a psychologically distinctive sense, and I explore the ways in which the mechanisms supporting this relation have proliferated and changed over time.
Mar 2, 2018
The ability to reason logically is often taken to be a sign of language-like or conceptual thought. One way to investigate whether animals are conceptual thinkers, or have a language of thought, is thus to investigate their ability to reason logically. The Greek Stoic philosopher Chrysipuss reported anecdotal evidence that dogs reason by way of the disjunctive syllogism. More recently, animal researchers have provided supporting evidence from controlled studies. But philosophers such as José Luis Bermúdez and Michael Rescorla have appealed to forms of “proto-logical” reasoning to ground alternative explanations of this evidence. These forms of proto-logical reasoning are so powerful that they generate a challenge: How, if at all, can genuine logical reasoning be detected and distinguished from its proto-logical rivals? My talk will offer some suggestions.
Jan 26, 2018
Researchers often distinguish between two kinds of norms. On one hand, there are purely statistical norms (which specify what most frequently happens). On the other, there are prescriptive norms (which specify what ought to happen). In this talk, I discuss a series of new studies we have conducted on people’s ordinary way of thinking about norms. These studies provide evidence that people actually integrate these two kinds of norms into a single undifferentiated notion of the “normal.” Subsequent studies have explored the implications of this idea for a number of different questions, including issues in semantics (e.g., intuitions about gradable adjectives) and issues in cognitive science (e.g., the way people ordinarily think about alternative possibilities).
Paula Droege & Victoria Braithwaite
Mar 24, 2017
Pain is an enigma, we all know what it is, but nobody other than you can experience the pain you feel. Because animals cannot tell us their experiences, it is difficult to understand whether they experience pain and if they do, how they process it. To address this problem, we need to consider the mechanisms that underlie pain, and then investigate the capacity for these kinds of process to occur in different animals. Constructing lists of attributes that are normally associated with human pain – cognitive bias, plasticity of nociceptive response, long-term learning – is inadequate without a theory that shows how these attributes demonstrate the capacity for conscious pain. We believe the affective aspect of pain, the conscious feeling of aversion, is central to concerns about animal pain and welfare. While dissociable from the sensory aspect of pain, both are normally interconnected and adaptive. In our view, the felt emotion of pain aids the animal in flexibly responding to its environment and situations involving pain. Consciousness is necessary to determine a situation-appropriate response under novel conditions; to attend to and select relevant stimuli; and to tailor actions in pursuit of goals.
Dec 12, 2016
Animal communication is usually understood in terms of signalling: Senders emit signals that convey some information to receivers. Since information transfer appears to play a central role in explaining animal behaviour, there has been some debate about the kind of information involved. In view of general philosophical theories of information and content, animal signals are thought to either convey a kind of natural information or a kind of representational content. Here I approach the topic from the point of view of the information contents that scientists actually attribute to animal signals. I will argue that actual content attributions often pick out the conditions under which receivers acquired their dispositions for signal- and situation-specific responses (‘acquisition conditions’). This yields a notion of receiver-dependent content that is explanatory and widely applicable.