Dan Lloyd

O'Brien & Opie's vehicle hypothesis is an attractive framework for the study of consciousness. To fully embrace the hypothesis, however, two of the authors' claims should be extended: first, since phenomenal content is entirely dependent on occurrent brain events and only contingently correlated with external events, it is no longer necessary to regard states of consciousness as representations. Second, the authors' insistence that only stable states of a neural network are conscious seems ad hoc.

Connectionism—also known as parallel distributed processing, or neural network modeling—offers promise as a framework to unite clinical and cognitive psychology, and as a tool for studying conscious and unconscious mental activity. This paper describes a neural network model of the case study of Lucy R., from Freud and Breuer's Studies on Hysteria. Though very simple in architecture, the network spontaneously displays analogues of repression and hallucination, corresponding to Lucy R.'s symptoms. Salient elements of Lucy's conscious experience are represented in the model by the activations of neuronlike processors in a fully interconnected network, without hidden units. The model learns to associate elements of experience that were associated in Lucy's case history. Some of these configurations of elements were traumatic for Lucy; trauma is modeled in the network by "emphatic learning," learning accomplished at an abnormally high learning rate. The model suggests that changing associations among conscious elements are sufficient to generate the symptoms Freud observed: apparent repression, hallucination, and recovery through therapy. In the case of Lucy R., Freud's theoretical inference regarding active but unconscious thought is not required by his data. Instead, the unconscious can be understood as a set of complex dispositions embodied in connections between elements of conscious experience.

In lieu of an abstract, here is a brief excerpt of the content:Commentary on “Searle and the ‘Deep Unconscious’”Dan Lloyd (bio)Can another person know my thoughts with better authority than I know them myself? With his affirmative answer to this question, Freud invented the twentieth-century human, a being whose mind is accessible to scrutiny from outside, and whose attempts at conscious self-explanation are at best partial and in many cases wrong. Even as Freud’s scientific influence wanes, the shift of authority he inaugurated is still gathering strength. Now the mind is open to many outsiders. Not only psychoanalysts, but behaviorists, cognitive scientists, and neuroscientists are all striving to read my mind, and yours. Against this historical sea change stands the solitary John Searle, whose work is consistently animated by fierce adherence to first person authority and surly attacks of any theory that could be exploited to undermine that authority. To most readers Searle’s positions seem stubbornly anti-scientific and covertly dualistic, despite his insistence to the contrary. And since science is good and dualism is bad, Searle’s arguments have become standing obstacles for theorists to discuss, debate, or dismiss.Buttressing his insistence on first person authority, Searle defends a Cartesian view of mental representation or intentionality, holding that an intentional entity is a conscious entity, either in actuality or potentiality. Deeply unconscious representations are therefore impossible. Eric Gillett has joined the debate just here with a pointed critique of Searle’s argument (also discussed in Lloyd 1990). Gillett performs a double service by challenging Searle’s premises and offering some positive grounds for the deep unconscious. I agree with Gillett that establishing the existence of deeply unconscious representations requires a theory of intentionality that is independent of conscious awareness. Ruth Millikan’s evolutionary teleology is an excellent candidate, in its pursuit of real function and real content (vs. “as-if” interpretation) for mental items.Some questions remain. First, if deeply unconscious entities are clinically important, then in addition to a defense-in-principle of their existence we will need some method of identifying specific deep representations in particular cases. Millikan’s functionalism links content to broad biological and environmental contexts, and thereby seems to be an unwieldy tool for content ascription. If deeply unconscious representations get their content in the way Millikan describes, how can the clinician make use of them?Second, and more important, Millikan’s theory shares with many others the ambition to explain all intentional entities, conscious and unconscious alike. This leads to a new dualism of mental contents, a split between intentional content and conscious content, since the comprehensive theory of intentionality will assign mental content both [End Page 201] shallow and deep, irrespective of consciousness. This is a perplexity that transcends particular theories of intentionality, since what my mental states represent seems at least partly dependent on my environment or context, but my current state of consciousness is exclusively dependent on my brain. How, in theory, can these two concepts of content be reconciled?In sum, deeply unconscious representations do not come cheap. One way to escape the new dualism of content is to deny that the deep entities are representations. Instead, they might be dispositions or the “mechanisms, instinctual drives, structures, and energies which can never become conscious under any conditions,” as Gillett suggests. (In Lloyd 1994 I explore the possibility of a dispositional unconscious understood via connectionism.) On this view, mental content could remain within our first person authority, while the causal processes that manage that content remain out of sight. I wouldn’t have to guess what I’m thinking, but I would depend on the sciences of the mind to tell me how I come to think it.Related ArticlesFeature Article: Searle and the "Deep Unconscious"Commentary: Commentary by LloydDan Lloyd Dan Lloyd, Department of Philosophy, Trinity College, Hartford, CT 06106, U.S.A.ReferencesLloyd, D. 1990. Loose connections: Four problems with Searle’s argument for the ‘Connection Principle.’ Behavioral and Brain Sciences 13: 615–16. Google ScholarLloyd, D. 1994. Connectionist hysteria: Reducing a Freudian case study to a network model. Philosophy, Psychiatry, & Psychology 1, no. 2: 69–88. Google ScholarCopyright © 1996 The Johns Hopkins University Press...

  Connectionism and phenomenology can mutually inform and mutually constrain each other. In this manifesto I outline an approach to consciousness based on distinctions developed by connectionists. Two core identities are central to a connectionist theory of consciouness: conscious states of mind are identical to occurrent activation patterns of processing units; and the variable dispositional strengths on connections between units store latent and unconscious information. Within this broad framework, a connectionist model of consciousness succeeds according to the degree of correspondence between the content of human consciousness (the world as it is experienced) and the interpreted content of the network. Constitutive self-awareness and reflective self-awareness can be captured in a model through its ability to respond to self-reflexive information, identify self-referential categories, and process information in the absence of simultaneous input. The qualitative feel of sensation appears in a model as states of activation that are not fully discriminated by later processing. Connectionism also uniquely explains several specific features of experience. The most important of these is the superposition of information in consciousness — our ability to perceive more than meets the eye, and to apprehend complex categorical and temporal information in a single highly-cognized glance. This superposition in experience matches a superposition of representational content in distributed representations

A philosophical zombie is a being indistinguishable from an ordinary human in every observable respect, but lacking subjective consciousness. Zombiehood implies *linguistic indiscriminability*, the zombie tendency to talk and even do philosophy of mind in language indiscriminable from ordinary discourse. Zombies thus speak *Zombish*, indistinguishable from English but radically distinct in reference for mental terms. The fate of zombies ultimately depends on whether Zombish can be consistently interpreted. If it can be interpreted consistently, then zombies remain possible, but no test could ever reveal whether anyone is speaking Zombish. Any materialist theory of consciousness is therefore already a theory in Zombish, and is equally confirmable in its human language edition and its zombie-language edition . On the other hand, if Zombish cannot be consistently interpreted, then the zombies described in Zombish are logically impossible. Either way, the search for a materialistic theory of consciousness should be untroubled by the zombies among us

For more than a century the paradigm inspiringcognitive neuroscience has been modular and localist.Contemporary research in functional brain imaginggenerally relies on methods favorable to localizingparticular functions in one or more specific brainregions. Meanwhile, connectionist cognitive scientistshave celebrated the computational powers ofdistributed processing, and pioneered methods forinterpreting distributed representations. This papertakes a connectionist approach to functionalneuroimaging. A tabulation of 35 PET (positronemission tomography) experiments strongly indicatesdistributed function for at least the ''medium sized''anatomical units, the cortical Brodmann areas. Moreimportant, when these PET experiments were interpretedas distributed representations, multidimensionalscaling revealed a ''brain activation space'' with asalient structure organized primarily by the sensorymodality of the stimulus, and secondarily by the typeof motor response. These results suggest that currentanalytical techniques in functional neuroimagingshould be augmented by distributed processinganalyses, and that these analyses may lead to manydiscoveries about the structure of ''inner space.''

In lieu of an abstract, here is a brief excerpt of the content:Commentary on “Towards a Design-Based Analysis of Emotional Episodes”Dan Lloyd (bio)To think about grief is to think about many things. My one-year-old daughter was practicing opening and closing a cabinet door as I puzzled over a response to Wright, Sloman, and Beaudoin’s “Toward a Design-Based Analysis of Emotional Episodes.” She was completely absorbed in her project, and as I watched my elf at her task, I thought about the voices of loss that speak briefly in section 3.1. How terrible it would be to lose a loved one. With that, I was overcome with a pang of love for my daughter, and further thought about the paper ceased.In the authors’ terms, my feelings about my daughter were “perturbances,” states that interfered with the internal management of my thought process, derailing it from the deliberate project of responding to the paper. Reading and thinking about loss activated the attachment structures involving those whose loss would be most devastating to me. The activation of the attachment structure was my episode of intense feeling, arising in me with an insistence level that temporarily exceeded my academic pursuits. The whole sequence of mental events arose through the interaction of major processes of perception, beliefs, “motivators,” plans, self-monitoring, and global control. The mind, according to the authors, is a collection of such processes.Is that all there is? The familiar objection to this brand of functionalist analysis is that it leaves out the most important part of mind, namely, conscious experience. Such an “architecture” for an “intelligent agent” is at best a simulation of some aspects of cognition, but the feelings, the moods, the subjective what-it-is-like of experience are missing. And without them, say the familiar critics, the mind is missing. The authors are naturally aware of this objection, and are careful to hedge their project: “In our discussion the ‘phenomenological problem’ is factored out and placed to one side.”I wonder, however, whether the authors are too quick to concede that their model omits the phenomenal. The symptoms of grief expressed by its sufferers would all be considered reports of the experience of grief, aspects of its phenomenology. The authors use these reports to guide their analysis of the functional processes of grief, and their analysis in turn explains why the grievers experience their losses as they do. Functionalist psychologists usually base their analyses on strictly observable responses obtained in cognitive psychology experiments. In using first-person reports as their guide, the authors inaugurate a promising union of phenomenology and functionalism. After their treatment of grief, the left over and left out phenomenology is noticeably slimmer.What remains of grief is its pain. In the spirit [End Page 127] of the paper, I will close with a brief speculative extension of the authors’ approach, using it to examine the core of the experience of grief. Like physical pain, emotional pain is curiously resistant to description, and not even easy to remember. (One remembers that one was in pain, and that it was hateful, but such reminiscences are faint shadows of the experience itself.) Though its phenomenology is elusive, pain has a clear function. It is aversive. As the authors note, pain involves “dispositions to terminate or reduce some state or process.” In their terms, pain is a motivator with a very high insistence level. Motivators motivate plans, and an aversive motivator would principally motivate a plan to remove the cause of the pain. However, grief and other chronic emotional pains are marked by irremediability. There can be no plan that replaces the loss, only fantasy plans or reversions to infant strategies of last resort (like crying). Within the cognitive architecture proposed by this paper, emotional pain is thus characterized by a collision of two processes, a highly insistent motive meeting a vacuum where a plan should be. The unanswered insistence would make the collision hateful, while the absent plan would leave a state partly without intentional content, and thus hard to describe or remember. This breakdown, arguably, is the pain—an emergent state or states arising in just this sort of wrenching bind.The traditional critic will insist that the painfulness...

Our heads are full of representations, according to cognitive science. It might seem inevitable that conscious states are a type of brain-based representation, but in this paper I argue that representation and consciousness each form conceptually distinct domains. Representational content depends on context, usually causal, as shown by familiar cases in which context varies while brain states do not -- twin earth cases and brains-in-vats, for example. But these same cases show that conscious content does not depend on context. The vatted brain, for example, enjoys the same experiences as its in vivo counterpart. The structure of experience -- its parts and their distinctive characters -- is the dynamic structure of the brain, viewed "from within." I call this position methodological phenomenalism (MP), and consider its prospects as a foundation for a science of consciousness. I close with a consideration of MP on the subjective "character" of conscious states. Turning away from representation dissolves the perplexity of subjectivity, leaving hopeful prospects for the scientific study of consciousness

Good research requires, among other virtues,(i) methods that yield stable experimentalobservations without arbitrary (post hoc)assumptions, (ii) logical interpretations ofthe sources of observations, and (iii) soundinferences to general causal mechanismsexplaining experimental results by placing themin larger explanatory contexts. In TheNew Phrenology , William Uttal examines theresearch tradition of localization, and findsit deficient in all three virtues, whetherbased on lesion studies or on new technologiesfor functional brain imaging. In this paper Iconsider just the arguments concerning brainimaging, especially functional MagneticResonance Imaging. I think that Uttal is tooharsh in his methodological critique, butcorrect in his assessment of the conceptuallimitations of localist evidence. I proposeinstead a data-driven test for assessingrelative modularity in brain images, and showits use in a secondary analysis of fMRI datafrom the National fMRI Data Center(www.fmridc.org). Although the analysis is alimited pilot study, it offers additionalempirical challenge to localism