Beyond the mind-body problem: Seven testable criteria for consciousness.
There is no more important quest in the whole of science probably than the
attempt to understand those very particular events in evolution by which
brains worked out that special trick that enabled them to add to the cosmic
scheme of things: colour, sound, pain, pleasure, and all the other facets of
mental experience. --- Roger Sperry (1968, p. 723)
Bernard J. Baars (*)
The Neurosciences Institute
10640 John Jay Hopkins Drive
San Diego, Calif. 92121
Many recent brain findings are said to relate to consciousness, but controversy and uncertainty persist about consciousness in fields like memory, attention and perception. One difficulty is that many scientists seem to believe that consciousness as a scientific problem requires a solution to the metaphysical mind-body problem. Instead, this paper suggests a set of seven practical, testable criteria for evidence about consciousness. The criteria are inductive and deliberately side-step philosophical debates. They sketch a path toward understanding consciousness without first overcoming mind-body puzzles.
Taken together, the seven criteria set a high threshold for the validation of brain events underlying consciousness. They imply that consciousness involves a distinctive operating mode of the brain with its own functional architecture (e.g. Baars, 1988, 1993, 1998; Edelman, 1989; Tononi & Edelman, 1998). The criteria point to a core brain anatomy and physiology for sensory consciousness, going back at least 100 million years to early mammals. A key question is how to integrate local neuronal events that can be shown to underlie conscious contents with evidence for their global brain influences. The evidence today shows no support for the "gap" that is often said to exist between the brain and subjective aspects of consciousness.
In 1989 Logothetis and Schall published an experiment on the "Neural correlates of subjective visual experience" in the journal Science, one of the first occurrences of the term “subjective experience” in a major science journal in modern times. Logothetis and Schall compared the different effects of conscious and unconscious pictures presented to macaque monkeys on single-neuron activity in the visual brain. Their article heralded a marked increase in brain studies of consciousness, after many decades of neglect. While conscious perception has been explored since Newton and Helmholtz, the topic lost scientific respectability at the beginning of the 20th century. In the last few decades that has changed. Consciousness and its brain basis are increasingly treated as normal science (Baars & Newman, in press; see references). Today, new findings appear regularly in leading journals.
Yet there is still considerable confusion about the kind of evidence that bears on the problem. For example:
1. The sophisticated experimental literature on selective attention makes almost no reference to consciousness (e.g. Pashler, 1997). Yet we know from daily experience that when we pay attention to something we become conscious of it (Shiffrin, 1997). For the last fifty years every single subject in every selective attention experiment ever published has become conscious of every stimulus selected in the experimental task. It is not that attention researchers do not think about these questions; rather, they seem to perceive some basic conceptual difficulties in dealing with them.
2. Likewise, we are conscious of the "fleeting present" only to lose it a few seconds later. Faded events are believed to remain in working memory for some seconds, and episodic recall aims to bring them to consciousness again (Tulving, 1985; Baddeley, 1992). These points were well known to Ebbinghaus and William James. Yet many memory researchers are only beginning to rediscover the relevance of consciousness to their field.
3. Finally, we are conscious of perceptual events every second of the waking day. Yet since the heyday of behaviorism, perception researchers have maintained that their subject can be studied without reference to consciousness (e.g. Stevens, 1966).
These core fields of study --- attention, memory and perception --- present a great puzzle. They are clearly related to personal consciousness, as has been understood since William James, Ebbinghaus, and Helmholtz. Yet modern researchers have a marked tendency to avoid this reliable fact. This is odd, because scientists as a whole are not in the habit of evading a chance to study the most fundamental "constitutive problem" in their disciplines (Miller, 1986).
After hundreds of conversations with researchers over two decades, it seems to me that many are reluctant to address human consciousness because they do not want to confront the philosophical mind-body problem --- the traditional question of how the domain of conscious mental events relates to the world of brains and bodies. Mind-body questions are notoriously intractable, and to many scientists it seems more sensible to avoid the topic altogether. As Chalmers (1997, p. xi) writes,
Consciousness is the biggest mystery. It may be the largest outstanding obstacle to our quest for a scientific understanding of the universe. The science of physics is not yet complete, but it is well understood; the science of biology has removed many ancient mysteries surrounding the nature of life. … Even in the science of mind, much progress has been made. … Consciousness, however, is as perplexing as it ever was. It still seems utterly mysterious that the causation of behavior should be accompanied by a subjective inner life.
However, Searle (1994) suggests that the fault is not in our stars, but in ourselves --- that we are simply approaching the problem in a confused fashion:
The famous mind-body problem, the source of so much controversy over the past two millenia, has a simple solution. This solution has been available to any educated person since serious work began on the brain nearly a century ago, and in a sense, we all know it to be true. Here it is: Mental phenomena are caused by neurophysiological processes in the brain and are themselves features of the brain. … Mental events and processes are as much part of our biological natural history as digestion … (p. 1)
Indeed, Searle argues that standard mind-body arguments are actively pernicious; they tend to lead us astray, just as controversies about the invisible ether led physicists astray and arguments about the “vital essence” of living things led biologists to simply ask the wrong questions a century ago.
Seen from the perspective of the last fifty years, the philosophy of mind, as well as cognitive science and certain branches of psychology, present a very curious spectacle. The most striking feature is how much of mainstream philosophy of mind seems obviously false. … obvious facts, such as that we all really do have subjective mental states and that they are not eliminable in favor of anything else, are routinely denied by many, if not most, of the advanced thinkers in the subject. (p. 3)
Whether one agrees or not, there is no doubt that philosophical controversy continues to rage today, as it has for centuries about the same basic questions. In such a climate science cannot thrive.
0.1 The smuggled premise of dualism.
One reason why mind-body questions tend to mislead is that their very wording smuggles in dualistic premises. "What is the relationship between mind and body?" suggests a mind-body dichotomy, just as "When did you stop beating your wife?" suggests an answer by the very way the question is posed. Even Searle's answer that consciousness is a brain function implicitly accepts the premise that there are two separate worlds, one of conscious experience, the other of physical extension. Mind-body questions appear to be seductively simple and straightforward, but the result is always a snarl of contradictions. No answer is ever permanently settled.
Again, this situation is not without precedent in the history of science. The question posed by vitalism about 1900, “what is the essence of living things?” also seems simple and seductive, and leads to a morass of scientifically unworkable or premature problems. Biologists needed to reject the very question asked by vitalists in order to address testable questions in an orderly way, a path that ultimately led to modern evolutionary theory, DNA, and the biogenetic revolution we see today. Scientists must be extremely careful to avoid conceptual traps and only address testable questions that lead somewhere. The extreme repetitiveness of mind-body debates over the centuries suggests that they do not give us the right questions to ask. In a sense, scientists are in the position of the mythical Indian sage, who when approached by seekers with any question whatsoever, would simply wag his finger. The mistake is in asking the question at all.
This paper suggests that productive research on conscious experience can proceed without trying to solve mind-body questions in the form that they are usually asked. Instead, one can posit testable criteria, based on existing studies, that apply to evidence that is within reach today. This paper suggests seven practical criteria. They suggest a path for scientific development without metaphysical worries. This paper aims to begin a dialogue about the relevance and adequacy of such criteria.
This is a traditional strategy in science. Conceptual puzzles are not new. They appear routinely when scientists begin to address almost any fundamental question. Testable criteria were worked out for other scientific questions at a comparable stage of research, for example to clarify the properties of neurotransmitters or the molecular structure of proteins, even before specific instances of the criteria were discovered (Platt, 1964). They helped set the stage for discovery.
The criteria proposed here are meant to be inductive, modest and achievable. They aim to provide a sound foundation for evidence and theory. They are also helpful to deal with other demands for instant answers. For example, it is premature to ask for deep, conceptual definitions of consciousness, because such questions can only be answered by highly developed scientific theory. While we have useful theoretical proposals today, none can claim to be definitive (e.g., Baars, 1988, 1998; Edelman, 1989; Tononi & Edelman, 1998). The same point applies to robots and zombies. Asking whether robots can be conscious presupposes an answer to the deepest theoretical questions. We do not have those answers; to insist on them is to invite endless, unproductive debates. Finally, this same approach applies to the question of adaptive function. The idea that consciousness has biological functions is almost certainly true. But we do not even understand the adaptive function of established phenomena like “silent” DNA, sleep, dreams, neuronal coding and the like. Functional explanations tend to come long after basic evidence is collected.
Asking for instant answers to advanced questions is putting the cart before the horse. The proper scientific answer to premature questions is: “I don’t know.”
That does not mean, however, that plain facts about consciousness can be ignored, or that productive empirical research should be avoided. Consciousness is an extraordinarily interesting and important scientific problem, which deserves serious study just as much as other fundamentals. The challenge today is to work out a practical approach to the evidence, gather the facts, and develop theory as the data emerge.
Note that a set of testable criteria does not constitute a theory. Rather, it occupies an intermediate level between evidence and theory. The aim of this paper is not a theory of consciousness, but a prolegomena to any future theory.
0.2 Real consciousness.
A scientific construct called "consciousness" would not be convincing if it were unrelated to our own experience. While science does not use subjective experience as primary evidence, researchers in fields like visual perception have always taken their own experiences into account. Many scientists run themselves as subjects in pilot experiments or even as formal subjects (when experimenter bias can be controlled). They would be surprised if their own conscious experiences differed from those reported by others. When reliable differences are discovered, as in studies of brain injury or other animals, they can become hot topics of empirical investigation. A central intuitive question that drives such research is "What is it like to have blindsight, or to see the world like a macaque monkey?" This is of course the core question --- "What is it like to be a bat?" --- posed by the philosopher Thomas Nagel (1974) to characterize the question of subjective experience. These points suggest that genuine subjective experience is indeed the goal of scientific research on consciousness (Baars, 1998b). A set of testable criteria must satisfy our strong and reliable intuitions about the meaning of consciousness.
Insert Table 1 about here.
1.0 Criterion 1. Contrastive method. Does the study treat consciousness as a variable?
The traditional philosophical question is “What is the relationship between mind and brain?” The question proposed here instead is “What is the difference between two similar brain or cognitive events, when one is demonstrably conscious and the other is not?” One reason why mind-brain questions are so difficult to answer is that comparing conscious humans to imaginary zombies or conscious computers involves enormous leaps of faith (e.g. Dennett, 1991). In contrast, empirical studies work by choosing comparison conditions that are as similar as possible. It may be that mind-body questions lead to endless controversy not because they are fundamental, but simply because they are insufficiently constrained.
Many recent studies of consciousness compare similar conscious and unconscious events in the brain (for reviews see Baars, 1988, 1998; Baars & Newman, in press; see section 1.1). Such comparisons treat consciousness as an experimental variable. This "contrastive method" is fundamentally different from the philosophical tradition since Descartes (Descartes, 1992; Block, 1995; Nagel, 1974; Dennett, 1991; Chalmers, 1996). The Cartesian tradition only sees the conscious side, but provides no comparison conditions without consciousness.
1.1 The traditional study of sensory consciousness. Another long tradition in science deals with sensory qualities like color, size and location as variables vis-à-vis similar conscious experiences. One can compare a shade of red to a slightly different shade of red without mentioning the fact that they are both conscious. Indeed Stevens (1966) explicitly made the point that one does not have to mention the word “consciousness” to do sound psychophysics. When we compare conscious events to each other we do not treat consciousness as a variable: It is redness that is the variable, or perceived size or location. It is only when similar conscious and unconscious events are compared to each other that the issue of consciousness as such emerges clearly, as figure from ground. Perceptual consciousness can be studied only when it is compared to subliminal perception, inattentional blindness, unconscious sensorimotor processes, and the like (Rock, 1983; Merikle & Joordens, 1997; Goodale & Milner, 1992).
The same point may be made about standard topics like memory and attention. As long as the conscious aspect of these topics is not treated as a variable of interest, it can be avoided. Yet attention and consciousness are intimately related because humans become conscious of events we pay attention to; and in immediate memory, we are conscious of things in the "fleeting present" only to lose it a few seconds later (Edelman, 1989). Likewise, episodic recall aims to make the original event conscious again (Parkin et al, 1995). The conscious aspect of memory and attention can be avoided only at a price: It is the price of lost opportunities for exploring one of the fundamental "constitutive issues" of mind and brain science (Miller, 1986, p. 220).
To some it may seem obvious that any scientific construct must be studied as a variable. But for those who have thought for years about mind-body questions it is remarkably difficult to change the habits of a lifetime. Thus to many thoughtful people this point is not obvious at all. This is again not unusual historically. Treating gravity or the position of the earth as a variable was difficult for Newton’s contemporaries. Treating species as variable was difficult and contentious to Darwin’s generation. The idea of treating consciousness as a variable is a hard-won insight (Baars, 1986, 1988).
1.2 Conscious and unconscious processes are often comparable. A key assumption of this approach is that in many cases conscious and unconscious events are comparable. If we use the language of conscious knowledge, we must also speak of unconscious knowledge. The question of unconscious knowledge has been extremely contentious since the mid-19th century, when Helmholtz first proposed “unconscious inference” (unbewuesste Schluss) to explain perceptual “fill-in” phenomena like the blind spot (Helmholtz, 1962). This notion remained controversial until quite recently (e.g. Rock, 1983). Some writers still do not like the idea of unconscious mental knowledge today, but whatever we conceive conscious and unconscious events to be, they must be capable of controlled comparison. If there are no unconscious comparison conditions, consciousness cannot be treated as a variable. Thus we need two inferred constructs: consciousness and unconsciousness, each with its own observable index and its own testable properties.
1.3 Evidence that conscious and unconscious brain events are often comparable.
Are there demonstrated cases in which similar conscious and unconscious brain processes can be compared? The evidence for that is now extensive (e.g., Baars, 1988; 1998; Baars & Newman, in press). For example,
1. Binocular rivalry. When different pictures are presented to the two eyes only one becomes conscious. Yet the unconscious input presented to the nondominant eye continues to be processed well into visual cortex. Early in visual cortex there is little difference between the two, judging by single-cell studies of feature neurons. However, object-recognition areas late in the ventral visual stream are apparently only triggered by conscious visual input, and not by matched unconscious stimuli (Logothetis & Schall, 1989; Sheinberg & Logothetis, 1997; Crick, 1996). These studies systematically apply the strategy of treating visual consciousness as a variable, keeping the content of the conscious and unconscious input the same. They currently provide some of our best insights into the cellular level of conscious visual brain events.
2. Inattentional blindness. One can look at an object without being conscious of it (Mack & Rock, 1998). An everyday example is reading a printed page while thinking of something else. At the end of the page we may suddenly realize we have been completely unaware of the meaning of the printed words. Subjects can look at a stream of superimposed pictures and words at the center of visual gaze, and be conscious of only pictures or only words. Using this technique, Rees et al (1999) have shown that the conscious focal elements activate corresponding specialized regions of cortex, while unconscious stimuli, even at the center of visual focus, do not evoke high level cortical activity.
3. Subliminal or backward-masked stimuli can be compared to conscious stimuli (Marcel, 1983). In vision, such unconscious stimuli reach the retina without signal degradation and may be processed well into the visual tract (e.g., Macknik and Livingston, 1998).
4. The "40 Hz phenomenon." A series of finding show that conscious contents correlate well with electrical activity in the gamma range (25-70 Hz). These results come from comparing conscious and unconscious brain activity for reversible figures, surgical amblyopia, or global state changes like waking vs. deep sleep and general anesthesia (Rodriguez et al, 1999; Kleinschmidt et al, 1998; Engel et al, 1999; John et al, in press). This evidence also depends on comparisons of similar conscious and unconscious events under well-controlled conditions.
5. Implicit knowledge in brain damage. Many types of brain damage allow comparisons between normal visual experience and the implicit visual knowledge that often remains after lesion. These include cortical blindness, parietal neglect, specific visual deficits and the like (Weiskrantz, 1991; Cowey & Stoerig, 1995; Goodale & Milner, 1992).
6. Brain electrical activity of the state of consciousness can be compared to sleep, coma, general anesthesia and epileptic states of absence, all unconscious states characterized by slow, widely synchronized electrical field activity. Waking and dreaming, on the other hand, show fast, desynchronized electrical field activity (Dement & Kleitman, 1957; Llinas & Pare, 1991; Steriade, McCormick & Sejnowski, 1993). Destexhe et al (1999) have shown that these large-scale field effects reflect interactions between billions of individual neurons in the thalamocortical system, with low correlated firing in conscious states, and highly correlated burst-pause alternation in unconscious states. Tononi and Edelman have provided a theoretical framework for these fundamental differences (1998), and their neurochemical basis is now increasingly well understood (Hobson, 1997).
7. Automaticity and habituation. Conscious task elements tend to become less conscious with repetition, providing another opportunity to treat consciousness as a controlled variable (LaBerge, 1980; Haier et al, 1992; Shiffrin, 1996; Raichle, 1998; Bargh & Chartrand, 1999). In general, with habituation a drop in cortical activity is observed along with an increase in basal ganglia activity, consistent with the notion that procedural automatisms are controlled by basal ganglia.
8. Memory research shows many systematic differences between conscious and unconscious processes. We are more conscious of episodic storage and retrieval that of semantic memory (Tulving, 1985). Recent evidence shows that retrieval processes are more conscious in recall than recognition (Parkin et al, 1998). The closely related distinction between implicit and explicit memory processes is pervasive (Schacter, 1990).
9. The unconscious contextual framework for focal consciousness. All conscious events occur within a framework, whether spatiotemporal (for perceptual contents) or semantic (for focal beliefs, for example). These “contextual” kinds of knowledge are largely unconscious when the focal contents for which they set the stage are conscious (Baars, 1988). For example, visual cells in parietal cortex do not evoke conscious experiences; yet they seem to set the spatial context for visual object cells located in the ventral temporal lobe which apparently underlie conscious object perception (Goodale & Milner, 1992; Sheinberg & Logothetis, 1997).