A key challenge for evolutionary science is to provide an account of the evolution of consciousness. While that is widely recognized, evolutionists have also been socialized from their very first moments in the field to believe that a defining feature of their approach to life is that evolution itself is not, has not been, and cannot be conscious. Taken together, this leads to an anomaly: On the one hand, evolutionists recognize and celebrate the central importance of the evolution of consciousness within the story of life, and on the other hand, most evolutionists deny its importance to the understanding of their own field.
For evolutionary science to play a role in society that takes full advantage of its enormous scientific precision, scope, and depth, this anomaly has to end. The reasons for this unfortunate clash of concepts are multiple, but they are all outdated and artificially limiting. Evolution can be, has been, and is consciousness—not in the cartoon forms imagined by a lay public, but rather as an emergent understanding central to a multi-dimensional, and multi-level extended evolutionary synthesis.
The etymology of the word conscious points to its central quality: these are actions that occur “with knowledge.” Stripped to the bone, consciousness can be thought of simply as the ability to respond to oneself and the environment and the regularities within and between them. When human and nonhuman animals show a dramatic diminishment of such responding, such as during sleep or comas, they are said to be “semi-conscious” or even “unconscious.” In a similar but more incremental way, as life forms evolve increasingly elaborate ways of responding to the external and internal environment and its regularities, in such forms as sensation, perception, and learning, they are said to become more conscious of their reactions and their surroundings. An organism that cannot show habituation due to repeated stimulation is less conscious of its environment than one that can; an animal that can detect and respond to antecedent-action-consequence regularities is more conscious than one that cannot.
It is difficult to imagine a world in which consciousness, so defined, is not a phenotypic result of evolution. That is so because of this bedrock fact upon which evolution itself is constructed: No structural or behavioral phenotype is successful in all contexts and thus context sensitivity will generally be useful. Resource acquisition, resource utilization, reproduction, protection of offspring, niche construction, niche selection, predation, avoidance of predation, avoidance of illness or injury, and so on can only be understood based on the selective features of the particular environments in which particular phenotypic variations occur and are inherited through genetic, epigenetic, behavioral, cultural, and symbolic means. The selective power of environmental fit means that the evolution of greater sensitivity to relevant environmental features internally and externally is virtually assured to be a core product of evolution itself. Consciousness, as I’ve defined it, will thus not only evolve; it is a key characteristic of the fitness of complex evolved systems.
Consider a life form that is better able to detect the presence of a predator due to heritable variations in its visual system. It is entirely correct to say that such a life form has evolved to be more conscious of the presence of a predator. The definition I offered earlier is fully satisfied: visually detecting a predator is based on heritable variations in response to light and its regularities, and the result is increased fitness. Other than churlish arguments over word choice, it is an empirical fact that consciousness evolved.
But was evolution itself conscious in such a case?
That is a tricky question because saying “yes” seems agentic. Variations are blind, or so we are told, and thus while the heritable changes in the visual system created a relative advantage in avoiding predation and as a result became more frequent, it was not purposive. The original change was blind. The visual system did not change in order to detect the predator.
This is only partially true (or I could just as easily have said “that is partially false”) because responding in order to produce particular visual effects is indeed part of the story behind the evolution of the visual system. Let me explain.
In the gene-centric era of evolutionary science, it would be easy to miss key features of the complex multi-dimensional and multi-level system that actually gives rise to a successful visual system. A well-adapted visual system requires more than a genetic capacity—it requires properly arranged developmental processes that foster phenotypic development such as peripheral and central nervous system stimulation, growth, and coordination. If a kitten’s eyelids are sewn shut during key developmental periods, it will never develop a normal visual system, even if the duration of the visual deprivation is only a matter of days.1 Note that behavior itself could result in poorly arranged developmental sequences much like this if behavior linked to vision were not constrained. For example, a kitten could in principle close its eyes too much, or hide its face in its mother’s underbelly for much of the day, or stare at the sun for hours on end—all of which would interfere with the proper development of the visual system. A healthy kitten does not normally do so because visual development is impacted by patterns of sensory reinforcement as part of a multi-dimensional system. For example, animals will work to avoid excessively bright lights or to produce positive changes in visual stimulation by head turning, exploration, or working to remove visual obstacles.2 Sensory preferences of this kind are of such importance that they enter even into animal rights conversations.3 In other words, evolution created sensory preference patterns that ensure that operant learning processes can play their small but important role in fostering healthy sensory and perceptual systems as part of a much larger system of distal and proximal sources of control over mechanisms of development.
Operant learning is purposive in a particular sense: changes in environmental contexts produced by actions in the past serve to alter the context for action now. Said in another way, operant learning is the past as the future in the present. This kind of learning affords a new kind of conscious, purposive behavior—responding in the present in order to produce something in the future that has been produced in similar situations in the past. In other words, it is a more elaborated form of consciousness based on an ability to respond to particular contingent regularities between environment and behavior.
Operant learning impacts other evolutionary processes such as niche construction and niche selection. Indeed, a good argument can be made that the Cambrian explosion was due to the evolution of operant and classical conditioning, which made it possible for organisms to seek out or to alter characteristic environments, changing the selection pressures that lead to speciation or other phenotypic developments.4 In that sense, it is fairly obvious that this form of increased consciousness altered the course of evolutionary development.
To some degree, all forms of evolved evolvability make a similar point. Bacteria that show more variation when placed in a growth medium that is missing key amino acids are showing a very limited form of “consciousness” that in turn will alter the course or evolutionary development.5But operant and classical conditioning are a clear leap forward—one in which the temporal and spatial features of an act in context alter how the environment impacts future actions.
If we grant that consciousness evolves and that consciousness impacts evolution, is it necessary to say that organisms evolve consciously? At least when we reach the level of consciousness represented by symbolic learning I think the answer is “yes.”
Human Symbolic Learning
By 12-16 months, a normally developing human infant who has learned that an object (say, a rubber duck) has a name (“duckie”) will orient toward the object when hearing the name, without specific training to do so. Furthermore, if the rubber duck squeaks, the infant will know that “squeak” is the sound “duckie” makes and vice versa even if the name and the sound have never been heard together.
Said in another way, an instance of one-way contingency learning (object → name) leads to a robustly two-way street of symbolic meaning that is then recombinable into symbolic networks (object ↔ name).
Deriving a network of the kind I have just described is called “stimulus equivalence” and although it is readily shown in human infants, after decades of trying, it has not been reliably produced in non-humans.6 Furthermore, we have known for more than 30 years that children who do not show stimulus equivalence do not develop normal human language.7
Stimulus equivalence marks a transition in the evolution of consciousness because it is the first example of a learning process that is relational, not associative.
Learned associations and direct acting contingencies are not robustly reversible or combinatorial. For example, in classical conditioning, providing food after a bell will lead to salivation to the bell, but not to a robust raising of ears when food is later presented. Backward conditioning is very weak and does not enter into long backward sequences when chains of events are provided (e.g., later presenting a foul odor before the bell may eventually lead to salivation at the odor, but not food avoidance based on backward associations with the odor). The reason backward conditioning is weak is that environmental regularities are not normally robustly reversible or combinatorial, and thus there is limited selection pressure to develop that open learning process. If an animal avoids predation by running to a thicket when it sees a lion, it does not mean it will avoid predation by running to a lion when it sees a thicket.
That lack of reversibility and combinatorial capacity is not true of relations. If I am certainly bigger than you, you are certainly smaller than me. The derived relation is just as robust as the known relation. The evolution of human language and cognition is based on this relational property.
From the beginning of the act of naming itself, some forms of relational learning are not limited to formal relations. In the context of a cooperative social group with some level of social referencing, joint attention, and perspective taking, regularities in naming can be made reliably reversible by paralinguistic or other cues. If this object is a “duckie” from the point of view of a speaker, then it can be entirely safe to assume within a given troop or band that a “duckie” is this object from the point of view of a listener. Relational terms like “is” demarcate this particular kind of cooperative regularity within a specific group.8
The human infant and toddler quickly learn to apply other reversible relations, increasingly controlled by arbitrary contextual cues. If a human infant hears an unfamiliar name it will search for an unfamiliar object in its environment and, if one is found, it will derive a two-way symbolic relation between the two.9 In other words, two relations of “different than” (the name is different than other names; the object is different than other objects) leads to a two-way “same as” relation (unfamiliar name ↔ unfamiliar object). As additional relations are added (comparisons, such as more / less; opposition, such as hot/cold; contingency, such as if → then; person, such as I/you; etc.), vast cognitive networks can emerge from very limited environmental inputs.
There is expansive experimental literature on this topic under the rubric of Relational Frame Theory that shows the ontogenetic histories needed to reveal these evolutionarily prepared responses.10 The claim I am making is that relational learning is the central core of human language and cognition, and evolved as an extension of cooperation.11
Evolutionists have noted that humans are particularly adept in relational learning tasks.12 In non-arbitrary contexts, these are defined by the relata themselves (e.g., a nickel is larger than a dime). What happens in symbolic behavior is that particular relational responses (e.g., “larger than”) are abstracted and then brought under the control of social cues, not just the related events (such as being told that a nickel “is smaller than” a dime). That “relational frame” allows any event to be related in any way to any other event by social attribution, and then to enter into larger and larger derived symbolic networks. For example, a first grader can be told that a penny is smaller than a nickel and that a nickel is smaller than a dime, and derive that a dime is larger than a penny. A three-year-old could not. Relational framing is evolutionarily prepared but also learned.
Relational learning of this kind is the smoking gun – the sine qua non of human language and cognition. We know that in part because children who do not show this kind of learning show only limited verbal and intellectual abilities, and whereas if they develop this kind of learning, they begin to advance more rapidly.13 This suggests that the unit of symbolic learning is relational, not associative.
Impact of Human Consciousness on Evolution
Symbolic learning is another step forward in the evolution of consciousness because with this repertoire of relational responding we can respond to the past as the symbolically constructed future in the present. Only a rather small set of cognitive relations are needed to solve problems through symbolic reasoning: names of events and their features, if → then relations, and comparisons. Stated more simply, human verbal problem solving involves an “if/then/better” relational network that alters present action so as to coordinate with the verbally constructed future. Responding of this kind is not only conscious, it allows symbolically intentional behavior.
The two-way street of human cognition transforms the present based on cognitive networks about the future. The evolving future that is presented symbolically in present moments via human language can alter the impact of the environment. Nelson Mandela can treat a prison guard kindly, for example, because that action brings a just world a little bit closer, even if the guard is a source of deprivation.
Said in another way, human cognition can change the “selection criteria” for human behavioral and cultural evolution. Genetic evolution depends on life and death. Human behavioral evolution does not remove that truth but supplements it with cognitively available meaning and purpose.
When people consider their future and apply evolutionary scientific concepts to actions and policy choices to alter that future, the world is consciously evolving. I believe that is a factual statement, but it is also pragmatically and politically useful to say that evolution can be conscious in that way because it provides a use for evolutionary science that will alter the receptivity of the public to this entire area of science.
Only a minority of the US population believes that human beings are as they are due to natural processes of evolution. I can’t help but think that is in part because evolution has not yet been shown to matter to the average Joanne or Joe. For that to change, evolutionists themselves need to show that they can solve problems of human concern. But for applied evolutionary science to emerge as a field, it is necessary to step up to the idea that evolution can be conscious, and then to spend much more time on the role of human behavior in evolving the future. The culture at large will not attend to evolution in a major way, in my opinion, until it is clear that humanity has the capacity to evolve on purpose, culturally and within a lifetime.
Evolution begins with processes of blind variation and selective retention, but it does not stay there for the simple reason that evolvability itself evolves.14 The phrase “survival of the most evolvable” is far truer to the whole of evolutionary data than the hoary phrase “survival of the fittest.” Symbolic learning is key to human consciousness, but human consciousness can comprehend and consciously apply multi-level and multi-dimensional evolutionary models to the accomplishment of human purposes.
Behavioral variation and selection within the lifetime of individuals is not merely an expression of genes and cultural practices. Learning is a legitimate evolutionary dimension that impacts on other evolutionary dimensions at other levels and time frames. Symbolic processes led to the principles of evolutionary science itself—variations within the relational networks of particular people were expressed and selected by accomplishment of their scientific purposes individually and culturally. If these principles then lead human beings to change their behavior in order to achieve better outcomes, and if the success of these actions maintain them—as would be the case with any successful application of evolutionary science that was sustained because of its utility —it seems impossible to avoid the conclusion that evolution can be conscious.
Applied evolutionary science is not just the passive beneficiary of scientific understanding—it is the very field in which an extended evolutionary synthesis will be fostered. We can think of applied evolutionary science as a type of fieldwork in the evolution of human behavior. No amount of laboratory knowledge is enough to be certain that the action of an organism is understood—but if this knowledge is applied in the actual environment in which the behavior occurs and predictable changes occur, the validity and utility of evolutionary science expands.
When we have created a robust field of applied evolutionary science, evolutionary science will be relevant to the world in a way that it is not now. And if applied evolutionary science is possible, it means that evolution itself can indeed be deliberate, intentional, purposeful, calculated, planned, and volitional. These are all merely terms for actions that are regulated by the “if / then / better” symbolic formations of human beings. Evolutionary principles can be applied to and contained by these formulations themselves.
We have evolutionary accounts of consciousness—now we need evolutionists to apply those accounts to their own assumptions, theories, and purposes. Understanding the evolution of consciousness provides the scaffolding for evolutionary science itself to consciously evolve, and to help human individuals and groups do so as well.15
Read the entire “Conscious Evolution” series:
- Can Evolution Be Conscious? Introducing a Collection of Commentaries Published on This View of Life by David Sloan Wilson, Mel Andrews, and Maximus Thaler
- Cultural Evolution, Insight, and Fundamental Theories of Consciousness by Liane Gabora
- Conscious Evolution is a Category Mistake by Massimo Pigliucci
- The Origins and Evolutionary Effects of Consciousness by Eva Jablonka and Simona Ginsburg
- The Evolution of Consciousness Enables Conscious Evolution by Steve Hayes
- Welcome to the Noösphere by Alice Andrews
 Hubel, D. H. & Wiesdel, T. N. (1970). Period of susceptibility to physiological effects of unilateral eye closure in kittens. Journal of Physiology-London, 206, 419-436. Doi: 0.1113/jphysiol.1970.sp009022
 An example of the motivational effects of visual variety on human infants is show by Caron, R. F., Caron, A. J., & Caldwell, R. C. (1971). Satiation of visual reinforcement in young infants. Developmental Psychology, 5(2), 279-289. Doi: 10.1037/h0031417. An example of the avoidance of intense illumination is shown by Taylor, N., Prescott, N., Perry, G., Potter, M., Le Sueur, C., & Wathes, C. (2006). Preference of growing pigs for illuminance. Applied Animal Behaviour Science, 96, 19-31. Doi: 10.1016/j.applanim.2005.04.016.
.Young, R. J. (2003). Environmental enrichment for captive animals. New York: Wiley-Blackwell.
 Ginsburg, S., & Jablonka, E. (2010). The evolution of associative learning: A factor in the Cambrian explosion. Journal of Theoretical Biology, 266(1), 11-20.
 Hersh, M. N., Ponder, R. G., Hastings, P. J., & Rosenberg, S. M. (2004). Adaptive mutation and amplification in Escherichia coli: two pathways of genome adaptation under stress. Research in Microbiology, 155, 353-359. Doi: 10.1016/j.resmic.2004.01.020
 A good initial review of that literature can be found in the book on relational frame theory cited in footnote 10 below.
 Devany, J. M., Hayes, S. C. & Nelson, R. O. (1986). Equivalence class formation in language-able and language-disabled children. Journal of the Experimental Analysis of Behavior, 46, 243-257. doi: 10.1901/jeab.1986.46-243
 For a more extended analysis of this idea, see Hayes, S. C. & Sanford, B. (2014). Cooperation came first: Evolution and human cognition. Journal of the Experimental Analysis of Behavior, 101, 112-129. doi: 10.1002/jeab.64
 Lipkens, G., Hayes, S. C., & Hayes, L. J. (1993). Longitudinal study of derived stimulus relations in an infant. Journal of Experimental Child Psychology, 56, 201-239. doi: 10.1006/jecp.1993.1032
 Hayes, S. C., Barnes-Holmes, D., & Roche, B. (2001). Relational Frame Theory: A Post-Skinnerian account of human language and cognition. New York: Plenum Press.
 See Hayes and Sanford, 2014 in footnote viii above.
 Penn, D., Holyoak, K., & Povinelli, D. (2008). Darwin’s mistake: Explaining the discontinuity between human and nonhuman minds. Behavioral and Brain Sciences, 31(2), 109-130. doi:10.1017/S0140525X08003543
 There is an extensive literature now of teaching relational framing skills to children with developmental disabilities, for example see Cassidy, S., Roche, B., & Hayes, S. C. (2011). A relational frame training intervention to raise intelligence quotients: A pilot study. The Psychological Record, 61, 173-198. These skills are known to be the bridge from simply saying a name in the presence of an object to being able to show higher levels of intelligent behavior: Belisle, J., Dixon, M. R. & Stanley, C. R. (2018). The mediating effects of derived relational responding on the relationship between verbal operant development and IQ. Behavior Analysis in Practice. Doi: 10.1007/s40617-018-0215-2
 Pigliucci, M. (2008). Is evolvability evolvable? Nature Reviews Genetics, 9, 75–82.
 Rather than tie down this paper with dense referencing, I have done so fairly lightly. The following references are particularly useful in exploring the arguments I am making:
Wilson, D. S. & Hayes, S. C. (Eds.). (2018). Evolution and contextual behavioral science: An integrated framework for understanding, predicting, and influencing human behavior. Oakland, CA: Context Press / New Harbinger Publications; and Wilson, D. S., Hayes, S. C., Biglan, T., & Embry, D. (2014). Evolving the future: Toward a science of intentional change. Behavioural and Brain Sciences, 34, 395-416. doi:10.1017/S0140525X13001593