My principal reason for revisiting and combining the essays in this volume was that they all converge on and develop a small number of basic themes. In the introduction, I try to bring those themes out and explain how the individual essays contribute to developing them. However, the process of editing the essays and thinking about how they connect together had a consequence that I probably should have foreseen. It made very clear where more work needs to be done. For that reason, I was very pleased when the referees for this volume suggested that I write an afterword identifying some of the challenges (and, hopefully, opportunities) that lie ahead in this area. I welcome being able at least to identify some of the gaps and problems that remain, even if I am not in a position at the moment to fill and solve them. But since this enterprise may seem a little self-indulgent, I will strive for brevity.
Several of the essays highlight the importance of a broadly Gibsonian approach to visual perception as revealing a form of nonconceptual self-consciousness in the structure of “ordinary,” outward-directed perception. These are also the earliest published essays in this volume. The philosophical discussion still seems to me to be sound, but there are several areas where it is plain that more work is needed. Chapter 1 briefly touches on the relation between Gibson’s ecological approach to visual perception and ongoing research in the neuroscience and neuropsychology of vision, particularly the “two visual systems” hypothesis. According to that hypothesis, most prominently developed by David Milner and Melvyn Goodale, vision is subserved by two systems that are both functionally and neuroanatomically distinct (Goodale and Milner 1996). Milner and Goodale distinguish between vision for perception (primarily concerned with object recognition and object identification) and vision for action (primarily concerned with how vision informs motor behavior).
Some aspects of Gibson’s idea that vision is simultaneously coperception of the self and the environment seem to fit uneasily within this bifurcation of vision. The essence of Gibson’s notion of an affordance, for example, is that we typically cannot separate out how we perceive and identify objects from our capacity to act on them. By the same token, self-specifying information in vision plays a structuring role in both identification and action. Moreover, as discussed in chapter 10 in the context of the Ebbinghaus illusion, Milner and Goodale suggest that vision for action can nonconsciously present information about objects that conflicts with how those objects are consciously perceived. If this is right, it has potential consequences for the idea that affordances and self-specifying information are part of the (conscious) content of visual perception. So, interesting and important questions remain at the interface between the neuroscience of vision, the ecological approach to visual perception, and the theory of content.
Answering these questions will hopefully shed light on another issue, which is that we need a clearer model of how self-specifying information features in the content of visual perception. It would be a fair criticism to make of my discussion of Gibson that I focus exclusively on making the case that the various types of self-specifying information are part of the content of perception, without providing a substantive account of what that content is in order to illustrate how they can feature in it. This is particularly important, given that the notion of nonconceptual content can be somewhat elusive. Certainly, Peacocke’s theory of scenario content (Peacocke 1992), which is the best-developed account of the nonconceptual content of perception, focuses (as currently developed) primarily on the exteroceptive (outward-directed), rather than proprioceptive (self-directed), dimension of vision. And, of course, a fully satisfying account of the nonconceptual content of perception would be, at a minimum, multimodal, incorporating touch and hearing (and, to a lesser extent, smell and taste), as well as vision. It would also, ideally, be integrated with a comprehensive account of the nonconceptual content of bodily awareness, in order to explain the representational grounds of intentional action.
Having such a substantive account of the nonconceptual content available would no doubt make it easier to elucidate much more clearly the complex epistemic relations between nonconceptual self-consciousness and conceptual self-consciousness. Some progress is made on this in the discussion of the epistemology of “taking at face value” in chapter 3, but there is much more to be done in elucidating the justificatory relations that hold between (self-) perception and (self-conscious) belief. From a personal perspective, I would hope that progress in this area might be made by integrating the discussion here of nonconceptual forms of self-awareness with the model of self-conscious thought and self-reference I developed in Understanding “I”: Language and Thought (Bermúdez 2017).
One of the principal themes of Understanding “I” is that the capacity for self-conscious thought (as manifested in, but not of course exhausted by, the ability to use the first-person pronoun “I” with understanding) depends on the ability to think of oneself as an object uniquely located in space and following a single path through space-time. Developing insights from Gareth Evans’s The Varieties of Reference (Evans 1982), I suggest that this ability can be understood in terms of practical capacities for self-location—for superimposing an egocentric understanding of space upon a nonegocentric cognitive map of the spatial environment. There are plainly affinities between that way of thinking about what it is to be a full-fledged self-conscious subject and the notion of a nonconceptual point of view developed in chapter 2—and, correspondingly, questions about how the notions relate to each other, both from a developmental perspective and from an epistemic perspective.
A further interesting avenue of inquiry concerns autobiographical memory. In Understanding “I,” I offer an account of why autobiographical memory has the immunity property (the property of being immune to error through misidentification: IEM) that emphasizes the relation between judgments based on autobiographical memory and the experienced episodes from which those judgments are derived. The basic idea is that a judgment based on autobiographical memory has the immunity property exactly when it is derived from an earlier experienced episode that at that earlier time either gave rise to, or could have given rise to, a judgment with the immunity property.1 In many cases, these experienced episodes will have involved states (e.g., perceptual states) with nonconceptual content. This raises the question of how nonconceptual states can ground autobiographical memories. Does this grounding relation simply duplicate that between nonconceptual perceptual states and perceptual judgments, or does the memory context add new dimensions?
There is a related issue here. In Understanding “I,” I emphasize the close connections between autobiographical memory and the capacity to take a narrative perspective on one’s one life. To have an autobiographical memory of an event is, in important part, to situate it within one’s personal history. As ordinarily understood, being able to think about one’s own personal history is a complex and sophisticated conceptual achievement—a paradigmatic form of linguistic self-consciousness. But, in line with the general approach of all the essays in this volume, we can ask whether this conceptual achievement emerges from, and/or is grounded in, a more primitive, nonconceptual way of experiencing oneself as existing over time. Is there, one might ask, a nonconceptual narrative self? And, if there is, what role does that nonconceptual narrative self play in underwriting and enabling our ability to conceptualize our personal history, to plan for the future, and to try to live up to the ideal of the person we want to be? Answering these questions will bring philosophical discussions of narrative and personal identity into dialogue with experimental studies of memory and the awareness of time in nonhuman animals and human infants.
The interface between philosophy and the cognitive sciences has been more systematically explored in the content of bodily awareness. But here too significant challenges remain. Several of the essays in this volume develop a model of the spatial content of bodily awareness—a model of how we experience the space of the body. This basic idea driving the model is that we experience the space of the bodily on a fundamentally non-Cartesian frame of reference. I am confident that this model will prove to be a powerful explanatory tool for understanding both normal and disordered bodily awareness, and the essays contain some programmatic suggestions in that direction. However, to make real progress, the model needs to be operationalized to bring it more clearly into contact with empirical studies of bodily awareness.
Chapter 8 takes initial steps toward fleshing out the model. It brings to bear an influential idea from kinesiology and robotics that the body can be modeled as rigid links connected by mechanical joints, and combines that idea with Marr and Nishara’s suggestion that all objects (and the human body in particular) can be represented as generalized cones (Marr and Nishihara 1978). But kinesiologists and roboticists are not typically interested in the phenomenology of bodily awareness. And Marr and Nishihara’s ideas about generalized cones were originally developed as part of a theory of vision, not of somatosensation. So plainly we still have considerable work to do in developing the model so that it can be brought to bear productively in experimental studies of how the body is experienced. Doing this will open up a range of experimental opportunities for exploring both normal bodily awareness and pathologies of bodily awareness in brain-damaged patients, as well as illusions of bodily awareness (such as the rubber hand illusion and the various whole body illusions).
The model developed here is a model of how the space of the body is experienced, which obviously falls short of a full account of the content of bodily awareness. It is one thing to say how we experience the location of a particular bodily event, and quite another to explain how that bodily event actually represents the ongoing state of the body. What sort of information about the body is carried within bodily awareness? How is it encoded? And how can it be integrated with information about the body derived from vision and the other exteroceptive sensory modalities?
Developing a full account of the content of bodily awareness that answers these questions will pay significant dividends in two different directions. In the first place, it will contribute to understanding both our capacities for planning intentional actions and our abilities to execute complex movements. (I mention these separately because it seems that they involve different types of bodily information—coarse-grained for action-planning and much finer-granted for motor control.) Second, it will help explain the source of our sense of ourselves as embodied creatures—as a special kind of physical object distinctive in ways brought out vividly by Merleau-Ponty, as discussed in chapter 5. Our experience of the body qua for-itself (to borrow Merleau-Ponty’s phrase) has multiple sources, but surely somatosensation, proprioception, kinesthesis, and the other forms of bodily awareness are among the most important elements.
Another important element in our experience of ourselves as physical objects of a highly distinctive kind must surely be our awareness of our own agency. This raises another set of important problems directly related to issues discussed in this volume. It has become common for philosophers and cognitive scientists to refer to a “sense of agency.” In one way, this can be perfectly harmless terminology—if the expression “sense of agency” is simply a shorthand for our awareness of our own agency. But it is sometimes interpreted as if there were some kind of qualitative marker of agency, a specific feeling that marks one out as the author of one’s actions.
The potential equivocation here closely matches that between different ways of understanding the putative “sense of ownership” discussed in chapters 6 through 9. In those essays, I outline and defend a deflationary account of the sense of ownership, accepting that there is a positive phenomenology of ownership but offering a reductive account of that phenomenology, as opposed to postulating a specific “feeling of mineness.” It seems likely that a similarly deflationary approach will prove profitable for understanding the sense of agency. Chapter 9 provides support for that thesis, particularly in the suggestion that agency provides a common thread between our sense of ownership for our own bodies and body parts (what I called φ-ownership) and our sense of ownership for our thoughts (ψ-ownership).
Turning from our awareness of our own agency to analyzing what that agency consists in, chapter 10 explores the relation between intentional action and commonsense psychology. It points to a range of phenomena suggesting, first, that there are ways of thinking about the springs of intentional action that do not involve the propositional attitudes, and, second, that we can navigate the social world and coordinate with other agents without engaging the explanatory framework of commonsense psychology. The obvious challenge here is to investigate how far these suggestions about the etiology and explanation of action can be scaled up, so that we have an accurate picture of the scope of commonsense psychology. There is a broad spectrum of possibilities. At one end lies the view, widespread in philosophy and in some currents of thinking in cognitive psychology (particularly the computational theory of mind as developed by Jerry Fodor) that “intentional action” is more or less synonymous with “action resulting from, and explicable in terms of, beliefs, desires, and other propositional attitudes.” At the other end lies eliminative materialism and the view that the explanatory framework of propositional attitude psychology is an otiose fiction. But there are many possible ways of steering between the two extremes.
At least three vibrant research programs in cognitive science are relevant to those projects of charting the scope and limits of commonsense psychology. One, already briefly considered above, is the “two visual systems” hypothesis, which has built into it the idea that the springs of action deploy two different systems, both functionally and neuroanatomically. The second explores social understanding in prelinguistic and nonlinguistic creatures. There has been extensive investigation of how well (if at all) nonlinguistic animals and young children understand psychological concepts, including propositional attitudes. Work on implicit false belief tasks (as proposed, e.g., in Onishi and Baillargeon 2005) has been taken to suggest that even young infants are capable of understanding basic psychological concepts such as belief. Likewise, experiments into the “theory of mind” of primates (and a range of other mammals) have been taken to suggest comparable metarepresentational abilities in nonhuman animals. As I have suggested elsewhere, however, these two research paradigms may be better interpreted as helping us to conceptualize alternative modes of social understanding that bypass the propositional attitudes (Bermúdez 2003, 2009a). Finally, it seems likely that the predictive brain research program, which treats the brain as a Bayesian hypothesis-testing machine (Hohwy 2013; Clark 2015), will yield powerful insights into how we can predictively model and navigate social interactions without deploying the cumbersome machinery of propositional attitude psychology.
This discussion raises a further challenge. The narrower the domain of commonsense psychology turns out to be, the more important it is to develop a model of rationality that can apply to action that is not driven by the propositional attitudes. Whether we are thinking about rational action and rational choice informally (through some version of the belief-desire law that rationality typically requires acting so as to best satisfy one’s desires in the light of available information) or more formally (through some version of expected utility theory, which is the touchstone theory of rationality in the social and cognitive sciences), the conceptual framework is provided by propositional attitude psychology. This is obvious for the belief-desire law, and follows in the case of expected utility theory if we think of utilities and probabilities as regimentations of desires and beliefs respectively (Bermúdez 2009b). So, if we whittle away at the connection between intentional action and propositional attitude psychology, and if we want to retain the idea that intentional action can be assessed for rationality, then we need either a new theoretical framework, or a new way of applying the old framework.
So, quite plainly, we have a more than plentiful supply of problems to solve and questions to answer. The bodily self is a rich topic, where progress will require cutting across many of the standard divisions of academic life. That is why it is so difficult. But this is also why it is so exciting. I plan to continue working on these problems and questions, and I hope others will too.