the fragmented mind
long-form dissertation abstract.
Taking this very string of words from the page and processing them together as a coherent whole, a thought, is remarkable feat of the mind. As you glance past each word they fade from your immediate view and yet you keep traces of them alive, in an instant combining them with prior knowledge to grasp the author’s meaning. Psychologists since James (1890) have attempted to model and understand this ability to build an active present, the bridge between our immediate past and the possibility of future action. This capacity to hold information in mind and to use it to guide our behavior is now at the heart of a foundational construct of cognitive psychology, working memory (Baddeley 2007). As philosophers incorporate scientifically vetted capacities to unravel longstanding questions and puzzles of the mind, working memory has begun to play a central explanatory role in new theories of “core” or “central” cognitive processes, including consciousness (Prinz 2012; Dehaene 2015), reasoning (Evans & Stanovich 2013; Mercier & Sperber 2017), and reflection (Carruthers 2015). Drawing on philosophical method and my own background in neuroscience, this dissertation challenges both the nature of working memory as the workspace of the mind and its role in explaining consciousness.
But why focus on working memory? Philosophers and psychologists both have a shared commitment to the notion that there is a center—the source of deliberation and reasoning, the core of personal identity, and the control structure for volition—in the mind. In philosophy, appeals to cognitive unity as a key factor in human mental achievements stretches back from Kant’s description of the rational agent (Korsgaard 2009), to the Cartesian meditator (Carriero 2009), reaching to Aristotle’s account of the virtuous agent, or phronimos, who deliberates with the aid of phantasia, our innate ability to form and retain images from past experience (De Anima 431a16). Doris summarizes this sentiment well in his recent work on agency: many philosophers “want to find a there there—a person, rather than haphazard muddle of cognitive systems.” (2015, 3). Working memory, borne of “cognitive revolution” of the 1960s and 1970s, was conceived as a general system situated between perception and action whose task is to keep information in mind and to use that information in the service of our goals—for instance, solving the 20th century problem of holding a phone number in mind long enough to write it down (Atkinson & Shiffrin 1971; Baddeley 2007). As working memory became embedded in the landscape of psychology, its purview increased to where it is now responsible for, as Jonides and Nee (2006) put it, “such acclaimed human intellectual achievements as reasoning, language-processing, and problem solving,” and general “thinking” and “planning” (223; Funahashi 2006, 251). As mentioned above, recent accounts of central cognitive functions have taken advantage of working memory and its large supporting literature to help scaffold theories aimed at resolving longstanding issues in the philosophy of mind.
This dissertation inverts this narrative. Using the analytic tools of philosophy of science, I review empirical developments surrounding working memory and conclude that there is no single-purpose system that can carry out all the functions that psychologists and philosophers have assigned to the “center” of the mind; that is, there is no there there. Instead, working memory collects and describes a plethora of fragmented systems and paths that creatures can use to get from perception to action. Deflating working memory also greatly problematizes many first-order, naturalistic theories of consciousness. While this fragmented view of cognition does imperil theories of consciousness and reasoning that rely on working memory, it also the creates the necessary space for socially-embedded, less-individualistic, less-anthropocentric, and anti- ableist accounts of these central processes that I hope to articulate as part of my larger philosophical project.
The dissertation consists of two parts. Chapters one and two survey the history of working memory, and argue that there is no single capacity that underlies all of the central processes associated with working memory. Chapters three and four apply the deflationary narrative to prominent first-order theories of consciousness. Finally, Chapter five critiques the concept of report on which these theories rely. As a general note to readers, I strongly encourage that they consult the index of arguments, the Lucius Partibus, that I have provided in the front matter as they peruse the main text. This dissertation sits at the border of theoretical neuroscience and philosophy of science, and as such I am compelled by norms of charity to provide an abundance of explanation on both fronts to make my claims and arguments as accessible to the broadest audience that may find them useful. As such, some sections—for instance an explanation of natural kindhood or a description of the basic properties of working memory models—may be redundant for some readers. Consulting the Lucius Partibus will prevent readers from missing or overlooking any important argumentative steps.
Chapter one reviews how working memory has transformed from a general system of cognition and learning (Atkinson & Shiffrin 1971), through Baddeley’s hugely influential sensory-specific “multicomponent” model (2007), to recent “emergent” accounts that view working memory as a process distributed throughout the brain, as opposed to a single-purpose system (Postle 2006; Christophel et al. 2017). This survey allows me to distinguish and define key terms that consistently appear in the literature, such as the “maintenance” and “manipulation” of information. Finally, I argue that maintenance is a version of manipulation, and that the broad scope attributed by these theories to working memory represents an error ossified at the heart of the construct; namely, that working memory is understood as a voluntary capacity that stands-in for most cognitive processes (Craik & Levy 1976).
In Chapter two, I argue that working memory is treated as a natural kind; that is, the properties associated with working memory, and how they are implemented in the brain, mirror a natural division, or “joint,” in the mind. This point is made explicit in Carruthers’ recent philosophical review of working memory (2015, 180). Homing in on the “maintenance” of information as the property common to a generic account of working memory, I argue that working memory does not satisfy even a congenial, or pluralist, account of natural kinds (Khalidi 2013). I show that maintenance of information occurs throughout the nervous system, and is associated with numerous processes not unique to working memory; for instance, even the retina can be understood to maintain and manipulate information when we perceive an afterimage (Shimojo et al. 2001). I close with a dilemma: either working memory is an overarching term that refers to the diverse systems that maintain and manipulate information, in which case it reflects cognition wholesale, or working memory refers to an artificially restricted subset of these systems, in which case it cannot explain all of our central cognitions. Neither horn secures working memory’s status as a natural kind—that is, as an objective division of the mind.
Chapters three and four begin the large exegetical project of examining how working memory is manifest across five prominent, first-order accounts of consciousness. Ultimately, I critique nearly all the theories for their over-reliance on a unitary and broadly capable model of working memory that does not exist. Chapter three reviews Prinz’s AIR theory of consciousness (2012). I argue that Prinz account cannot adequately separate attention from working memory and thus falls prey to the same charge of centrality that he lays against global workspace theorists (see arguments IV and V in the Lucius Partibus).
Chapter four then reviews four other first-order accounts, beginning with Baars’ original global workspace model (1988), Carruthers’ modern variant of the same model (2015), Lamme’s recurrent processing framework, and Dehaene’s global neuronal workspace theory (2014). The bulk of the chapter is a close examination of Dehaene’s theory, relevant findings from the neuroscientific literature, and a careful critique of his view. In particular, I argue that Dehaene’s four signatures of consciousness are neither necessary nor sufficient for consciousness, that the workspace cannot be manifest in frontal areas of the brain, and that his equation of the workspace with working memory threatens to render the theory trivial (see specifically arguments VIII, X, and XI in the Lucius Partibus). I also examine a serious concern I have with the structure of subliminal paradigms to hunt for the correlates of consciousness, arguing instead that these paradigms only detect signs of stimulus novelty or task-saliency (argument IX).
Chapter five presents a more theoretical critique of the concept of reportability. Report is taken to provide a window to conscious experience and sits at the center of many of these first-order theories of consciousness. I argue that it does not form a functional or anatomically unified kind, and that it is only unified by an implicit assumption that ties it to conscious experience. However, behaviors of “vegetative-state” patients challenge this assumption; for instance, consider that patients with anencephaly demonstrate a range of report-like behaviors. This forces a dilemma, particularly germane to global workspace theorists, as they must either revise a large portion of their theory to accommodate these behaviors as reports, or they must drop the implicit assumption that report is tied to consciousness. Doing so further requires that they provide additional criteria to differentiate bona fide instances of report.
Finally, I spend a few moments meditating on the purpose of the dissertation and the future directions of my philosophical project. Conceived as a palliative project, this dissertation does much of the necessary negative work—disabusing philosophers and psychologists alike of their reliance on conceptual black-boxes—that prevents the successful articulation of a more fragmented account of cognition, one that takes seriously our nature as social creatures. Ultimately, though the dissertation is negative, I do hold that we can keep information in mind, that we are conscious, and that we can reason, but how various animals or people will realize these abilities will differ and is not contingent on the possession or operation of a single-purpose internal “center” of the mind, most recently conceptualized as “working memory.” This project thus mirrors my own experiences and commitments as a scholar of both the humanities and the social sciences; by interfacing philosophy of science with empirical psychology and deeply considering how accounts of the mind shape our understanding of ourselves and creatures like us.
back to research.