Neuroscience Looks at Self-Consciousness


Self—the “I” we experience as our identities; in our bodies; with our histories and future possibilities; our personalities, our moment-to-moment experiences of ourselves, our environment, and relationships; our ability to decide and act; and our awareness of all this—is one of the great mysteries of philosophy and science.

A rich psychoanalytic literature has explored the psychology of the self since Heinz Kohut’s pioneering work a half century ago. Psychoanalytic self psychology is fundamentally relational: the child’s self is considered to emerge in the parent-child relationship, and self pathology can only be observed and treated by looking at its manifestations in relationships. In recent years, social neuroscientists have begun to look for underpinnings of this relational self in the biology of the brain; an example is Aikaterini Fotopoulou’s and Manos Tsakris’s recent argument of social contributions to the development of at least some of the body’s self-regulatory mechanisms.

A different approach has emerged from the study of consciousness. These researchers ask how the sense of a self with the ability to experience and act emerges from the activity of billions of connected neurons in the brain. The approaches are not incompatible: the kinds of phenomena the consciousness researchers are discovering must be the underpinnings of the self’s development in early relationships. But the neuropsychoanalysts are likely to be right—to study self- and consciousness-development without reference to social relationships may lead into blind alleys. Is it too early to ask if the biology of early relationships is a key to understanding consciousness itself?

As background, in this post I will look at a fascinating integrative review by Ryan Smith of the University of Arizona which looks at the neuroscience of the “experiential-agent” self. Smith is trained in philosophy as well as neuroscience. While he does not get deeply into the biology of brain regions and circuits, the overall concepts he discusses are an important structure for thinking about these problems.

The “experiential-agent” self Smith focuses on is “a unified entity with which we intuitively self-identify, and that (1) has access to all and only the information that we ourselves are conscious of and (2) uses that information to make deliberative decisions about how to act in the world.” It does not include such things as blood pressure regulation and automatic habits over which we do not exert conscious control.

Smith argues that this self’s operation takes place in the largely anterior brain system that “accesses conscious representations and uses them to make decisions;” in contrast, the self-concept is “represented within the brain’s distributed concept representation system.” Thus, the brain’s operations from which experience and the sense of agency emerge are one thing, and the idea of self-hood, which is a thought, occurs in a different brain system in which thoughts emerge. He is claiming, then, that “there is a system in the brain that ‘fits the description’ of the [emotional-agent] self-concept,” which is “distinct from the brain’s representation of that concept.”

This experiential-agent self is only one among many concepts of self. In some contexts we may think of ourselves “using a visual or proprioceptive image” of the body rather than in terms of experiencing or agency. Another view of the self is Antonio Damasio’s “core self,” which Smith describes as “made up of . . . ‘2nd order maps,’ or those conscious brain states that represent moment-to-moment body state changes due to interaction with the external world.” Smith notes Damasio’s claim that “the integration of this core self with episodic memory could give rise to an ‘autobiographical self,’ or a stable representation of one’s body as it has interacted with the world over time.” (Damasio’s bringing in interaction with the environment is one place these theories of self-consciousness may interact with psychoanalytic relational views of the self.)

Smith goes on to look at arguments against a neurally based experiential-agent self, beginning with the philosopher David Hume's 1739 claim that the self is “simply a bundle of momentary perceptions that the imagination happens to weave together.” The modern philosopher Thomas Metzinger argues that the brain generates a “useful representational illusion,” a “‘self-model’ to aid in certain adaptive computational functions. Michael Gazzaniga’s work on split-brain patients suggests that “belief in a unified self can be understood as a sort of plausible (but false) explanation for our own behavior, which is generated because this interpreter system does not ‘know about’ all the intricate and separate control processes in the brain that are actually involved.” Similarly, Daniel Dennett “suggests that the self may be more like the abstract ‘center of gravity’ of physics; that is, just as the center of gravity acts as a useful reference point. . . even though there need not be any physical thing that exists at that point [e.g., for hollow objects], our mental notion of self may similarly act as a useful narrative reference point in constructing our representations of the world.

All these arguments against the “reality” of an experiential-agent self in the brain are attempts to deal with the distributed nature of brain functions: it is now clear that “there is no single place where all of the information related to conscious experience is combined together and from which all behavioral commands are sent forth.” Recent models propose that “cortical and subcortical neural systems continually compete for control of action."

Smith’s defense of the experiencer-agent self rests on the idea that higher level neural mechanisms “can be understood in terms of interacting lower-level mechanisms nested within them.” He refers to Francis Crick’s and Christoph Koch’s distinction between posterior cortical regions which “represent perceptual features’” (Smith's aside: some cells in these regions “also activate when these same features are imagined in the absence of an external stimulus, or when memories of such features are retrieved”) and the frontal and tightly linked parietal regions “that selectively receive inputs from neurons representing percepts and memories, and which use these signals to update goals and inform the selection of appropriate actions.” He goes on, “Because this frontal-parietal system selectively takes representational activity as input and issues goal-directed action commands as output, it has been functionally characterized as being engaged in ‘deliberation,’ ‘decision-making,’ and ‘planning.’” Crick and Koch proposed “that it is useful to think of the front of the brain as ‘looking’ at the representations housed in the back;” the front of the brain is thus the “little man,” or “homunculus” that observes representational activity.”

Smith discusses the “infinite regress” problem of the homunculus: if we have in our heads a self that gathers our thoughts and perceptions, analyzes them, and decides on actions, what happens in the brain of that “little man,” and what about the brain of the little man inside the one inside that one? He notes that this problem disappears if the processes that generate experiential-action self are nonconscious and, importantly, simpler than the larger-scale process of self-conscious that emerges.

Smith goes on to reference research showing that “we are not directly aware of our own verbal thoughts, but only indirectly aware of them when they are re-represented in perceptual terms (i.e., when we ‘hear’ these thoughts spoken in our internal monologue).”

Neuroscientists have found several “neural signatures” of conscious experience: a specific pattern of increased frontal-parietal activation; the appearance on the scalp of an electrical potential called the “P3 wave;” a late burst of high-frequency oscillations within individual neurons in cortical sensory systems; and a large-scale increase in synchronous activity, signaling increased information exchange, between anterior and posterior regions. He says this implies that consciousness requires both the representation of information and its selective “broadcast” to and receipt by frontal-parietal systems that can use it to make goal-directed decisions.

The “global neural workspace” theory proposes that “conscious access is determined by a competition between several unconscious representations that are simultaneously active. . . Once one competing representation exceeds an activation threshold, top-down influence from frontal-parietal cortex is initiated that amplifies/maintains its activity, leading to the P3 wave, gamma bursts, and long-range synchrony described above. This. . . allow[s] the ‘message’ of the selected representation to be successfully received and used by a large number of cognitive processors simultaneously. . . leading to flexible use in goal-directed decision-making. . . while any single one of these cognitive processors might receive/use such information unconsciously, consciousness is said to occur when a representation is globally broadcast to all of them at once.”

Work on large-scale brain networks has identified “a set of highly connected cortical ‘hub’ regions (including frontal, parietal, cingulate, and insula regions) that govern long-range synchrony in the brain”; this overlaps with the global workspace network. “Each hub/processor can be activated alone in an unconscious manner, and have a resulting unconscious influence on behavior. What is conscious is only the content of the messages that these hubs/processor jointly receive and use as a result of global broadcasting, which can in turn allow for an ‘all things considered’ judgment in which the different claims on behavior made by each of these processors can be resolved into a single, unified, goal-directed decision regarding how to act.”

Smith notes evidence that all representations can be activated unconsciously—“images, words, numbers, meanings, values/rewards, errors, and actions.” Computational models suggest the brain performs many probabilistic calculations in parallel and samples the most probable hypothesis, constrained by the current context of goals and concerns.

To understand better how such a collection of unconsciously-operating processes can lead to a sense of conscious self, Smith uses the metaphor of a jury or other committee, in which several interacting members, each of which has its own opinions and perspectives, reach consensus decisions which are made available for action. He notes that there is no requirement that the elements be in the same physical place: “what matters is that the right type of causal interactions are in place to allow the committee . . . to act as a ‘causal hub’ or ‘control center’ where information is integrated and a decision process takes place.

Smith argues that the top level of control integrates conscious information about context, background knowledge, and current goals, while the lower, more perceptually based, levels are not sensitive to such factors. “While much of the time these lower levels of control can operate relatively independently (guiding action in habitual ways), the higher cortical level can ‘step in’ and take over whenever habitual responses conflict with current goals in a particular context. Though sometimes the highest level of control fails to intervene, and habit based responses (under subcortical control) occur that feel involuntary. Smith sees the top-level control mechanisms as “loosely coupled,” so they can exchange information and modulate each other’s activity.

Smith goes on to argue that the experiential/agent self that emerges from these interacting processes “does not have any direct experience of itself.” It has no access to its own internal workings. Self-awareness emerges via an indirect route, from conscious access to “representations of one’s own body, one’s own past memories, and one’s own perceived personality traits,” as well as the “self-concept stored within semantic memory systems. A question arises, then: if consious self-awareness amounts to an illusion assembled from these elements, how does consciously chosen action emerge from such an illusion? Smith does not appear to address this.

The nature of the interactions among the regions of the global workspace network are not understood in detail. Regions of the insula are “strongly implicated in representing and predicting the physiological state of the body,” “lateral prefrontal regions are implicated in representing goals,” and “anterior cingulate regions are known to play a role in managing response conflicts and predicting reward outcomes.” Smith suggests that the regions of the global workspace network may exchange “prediction and prediction-error signals during conscious processing until the network as a whole settles on a stable state that represents. . . the unified group perspective.”

Interestingly, Smith concludes with the suggestion that “unconscious self-external processes” do not “belong” to the experiential-agent self in any important sense. This would seem inconsistent with neuropsychoanalytic and social neuroscience perspectives of how the self develops during interaction with parent figures during infancy and early childhood. It would seem that while a great deal has been learned about self consciousness, and a smaller body of research supports the role of social interactions in the sense of self, these two perspectives have not yet been linked in any detailed consideration.

Further reading:

Ryan Smith, A neuro-cognitive defense of the unified self. Consciousness and Cognition 2017; 48:21-39.

Recent Posts