Conclusion

The main objective of my inquiry has been to assess the intuition that our minds are real, in the sense of being able to intervene causally in the course of natural events. I have constructed a conceptual framework that enables us to justify the conviction that cognitive properties are real. It is unavoidable to express myself in such a cautious way because we have seen that conceptual work is not sufficient on its own to establish the reality of cognitive properties. I will have achieved my goal if I have shown that it is conceivable that the mind is real in the same way as the human body or any other material object.

It might seem surprising to devote so much effort to justify such a modest conclusion. Indeed, nothing seems to be as certain to common sense as the reality of the mind. Similarly, in the philosophical tradition, the reality of the mind is often considered to be at least as certain as that of material objects. However, the enormous success of science in the systematic explanation of natural phenomena gives us solid reasons for considering the reality of matter to be established with certainty. More precisely, the doctrine that I have presupposed in this inquiry, and that has provided the metaphysical framework for my analysis, is “physicalism.” This term is preferable to “materialism” because it better reflects the rational basis for the choice of this metaphysical framework. That the various scientific disciplines, and above all physics as the fundamental science, are concerned with material objects and their properties is something that we have discovered empirically. It could have been that the scientific theories most successful in explaining and predicting phenomena are theories of mental phenomena. Physicalism is the metaphysical doctrine according to which science as a whole reveals what exists. The unification of the different scientific disciplines, which has played an important role in our thinking about the place of the mind in nature, gives physics a foundational place: scientific research gives us strong reasons to believe that the objects of sciences other than physics are composed exclusively of physical parts. The results of sciences such as chemistry and biology give us every reason to believe that chemical substances and living beings are composed exclusively of physical parts. Vitalist doctrines, according to which living beings also contain non-material components, such as élan vital or entelechy, have been refuted.

Physicalism results from the choice of science as a guide to ontology. Of course, science cannot refute skepticism; in other words, it cannot demonstrate the reality of the objects and properties that it investigates or of the laws that it discovers. Even less so can it demonstrate that what is not the subject of any science is not real. Finally, history shows us that the sciences are not immune to changes, sometimes radical, that lead to upheavals in the entities, objects, and properties in whose existence science gives us reason to believe. Therefore, it would be unreasonable to expect science as it stands today to give us definitive information about what exists. Certain substances, such as the phlogiston of medieval alchemy and the ether of pre-relativistic physics, have been eliminated from the set of entities for whose existence science is “ontologically committed,” to use Quine’s (1948) expression. Others have joined the set of objects or properties in whose existence science gives us reason to believe, such as quarks and the strong interaction force of nuclear physics, the olinguito (Bassaricyon neblina), a mammal discovered in 2013 in South America, or iconic memory, discovered during cognitive psychology experiments in 1960.

With regard to the particular limits of scientific knowledge, physicalism is a more cautious metaphysical doctrine than traditional materialism. Instead of adopting the dogmatic position that “everything that exists is material,” physicalism holds, first, that all objects that exist are either among the objects studied by physics or composed exclusively of objects studied by physics and, second, that all real properties are either among the properties studied by physics or reducible to them.

The debate about the reality of the mind revolves around the question of the reducibility of psychological laws. Property dualists and eliminativists maintain that these laws are irreducible. However, they do not agree on the conclusion to be drawn from this thesis. Dualists conclude that the irreducibility of psychological laws and the mental properties to which they relate shows that psychology is indeed an autonomous science that could never be replaced by neuroscience. Eliminativists conclude that the irreducibility of the mind demonstrates the radical falsity of psychological statements, at least as far as common-sense psychology is concerned. This falsity entails the non-existence of the properties to which its predicates seem to refer. Reductionists rely on the successes of cognitive neuroscience to argue that at least some cognitive abilities, such as long-term memory, are reducible to neuroscience and maybe even to biochemistry. Here again, however, it is possible to draw very different conclusions. Reductionists such as Schaffner (1993) and Bickle (2003) argue, against eliminativism, that the reduction of a cognitive property establishes its reality. But their conception of reduction leads them to deny that reduced properties have a reality distinct from the reducing properties: according to these authors, reduction shows that the reduced property is identical to the reducing properties. It is only insofar as the fixation of memory is identical to a neurophysiological (or biochemical property) that we are justified in considering it real. On the contrary, I have defended the thesis that the reduction of a property does not imply its identification with a property at the reducing level. In particular, I have shown that it is possible to reduce multi-realizable properties. Systems of different physical natures can possess a property such as temperature or the ability to learn by conditioning. Since different properties can give rise to a given emergent property, it cannot be identical to any of the reducing properties.

The whole debate on the reality of properties outside the realm of fundamental physics, and in particular on the reality of mental properties, presupposes a fundamental conviction that also motivates the adoption of physicalism. Dualism, eliminativism, and reductionism all agree in accepting the thesis that science is the ultimate judge of what is real, regardless of whether or not they support the identity of reduced and reductive properties. Thus, what justifies considering cognitive properties as real is the discovery of laws of nature that relate to these properties. Insofar as Rescorla and Wagner’s (1972) law describes a regularity in the dependence of the increase in associative strength between a conditioned stimulus and a response on different parameters — such as the salience of the unconditioned stimulus or the strength of association already present — we are justified in believing that learning by conditioning really exists as a cognitive process. The same applies to the properties of such learning, such as the increase in the strength of association as a function of the number of exposures to the stimuli.

The physicalist conviction rests on the observation of the progressive unification of the sciences. This is achieved largely through reductions between theories that study phenomena at different levels. The nature of reduction is controversial, and the first two chapters of this book were devoted to its detailed analysis. The model that I have developed is the result of a synthesis of two conceptions of reduction. From Nagel’s (1961) model, I have kept two theses. First, the reduction of a property involves deducing the laws that apply to it. For example, the reduction of temperature involves deducing the laws of thermodynamics that relate temperature to other macroscopic quantities, such as pressure or entropy. Second, this deduction necessarily involves laws of composition, which play the logical role of Nagel’s “bridge laws” or “linking principles.” Analysis of the reductions accomplished in the history of science shows that these laws of composition are not derived a priori from knowledge of the reducing level alone. The laws of composition are always constructed on the basis of prior knowledge of the two theories unified by the reduction. The laws of thermodynamics were not discovered on the basis of the laws of classical mechanics; they were first discovered on the basis of the observation of macroscopic phenomena: that is, in the context of research carried out at their own level independently of any consideration of the microscopic level. Similarly, the laws of learning by classical conditioning were first discovered by investigating the regularities observable at the level of cognitive systems — both animal and human — without regard to their microscopic neuronal and molecular constituents. It was only once the theory of macroscopic phenomena had been developed that the search began for regularities among the microscopic components of the objects of macroscopic theory — regularities that could serve eventually as premises in a deduction of macroscopic laws.

It might be objected that this is merely a matter of the contingent order of the acquisition of knowledge and does not contradict the thesis contained in the CHB reduction model (named after Churchland, Hooker, and Bickle), according to which the deduction of macroscopic laws can be achieved in principle merely by the conceptual analysis of a microscopic description. I have provided two responses to this objection. First, it seems to be convincing only insofar as ontology is not clearly distinguished from epistemology. On the ontological level, the physicalist framework of my investigation guarantees that macroscopic phenomena and laws are determined exclusively by microscopic states of affairs and the laws that apply to them. However, the difference between the CHB thesis and the thesis that the laws of composition are discovered only through knowledge of the laws at the macroscopic level is epistemic. Once this distinction has been made clearly, the applicability of the CHB model presupposes a possibility in principle that does not correspond to any historical reality: that of deducing a priori, from knowledge of the microscopic level alone, all of the laws of the macroscopic level. The thesis that this is possible in principle cannot be refuted directly; however, the burden of proof is on those who claim something to be possible that has never been achieved. Second, I proposed that there are cases of historical reductions in which the laws of composition include a part irreducible to the laws governing microscopic phenomena. The concept of an ensemble in the sense of Gibbs (1902) has no equivalent in the microscopic description of gas molecules. Without this concept — or others just as irreducible to the molecular level — it is impossible to deduce the macroscopic laws of thermodynamics. Macroscopic quantities such as temperature can be deduced only from the average of the squares of the velocities of all molecules. However, the average over time corresponds only to a real property of the system if that system is in equilibrium. Moreover, it is impossible to derive the fact that the system is in equilibrium from knowledge of the microscopic level alone.

My synthetic model of reduction also takes into account an important criticism of Nagel’s (1961) model of reduction. Research on historical reductions has shown that a reduction is generally accompanied by corrections to the reduced theory. These corrections are even the main motive for the search for reductions. To account for the difference between the theory shown to be deducible from the reducing theory, and the reduced theory as it was before the reduction, Schaffner (1967) introduced the concept of positive analogy: the theory T_R* that can be deduced from the reducing theory T_B is not identical, but only structurally analogous, to the reduced theory T_R. The CHB model takes account of the difference between the theory that needs to be reduced, T_R, and the theory T_R* that can be deduced from the reducing theory T_B. But the CHB model accompanies the recognition of this difference with a thesis that I have rejected: according to the CHB model, the deduction of T_R* from T_B is a case of intratheoretical deduction insofar as this deduction does not require any concept or principle external to the reducing theory T_B. The analogous relationship between T_R and T_R* then becomes the only intertheoretical part of the reduction.

According to the synthetic model of reduction that I have offered, it is generally necessary to use laws of composition that cannot be derived a priori from the reduction theory T_B alone. Therefore, contrary to the CHB thesis, the deduction of T_R* from T_B is not intratheoretical in T_B since it presupposes knowledge of T_R and sometimes requires recourse to new principles found neither in T_B nor in T_R. My synthetic model also recognizes — following Schaffner and the CHB model — that deduction from T_B and the laws of composition generally leads to a theory T_R* analogous to, but not identical with, the theory T_R that is the target of reduction.

The reduction of a systemic property is based above all on the discovery of a law of composition that determines that all complex objects with a certain structure necessarily possess this property. This discovery makes it possible to integrate the property into the system of scientific knowledge, the best way of justifying its reality. However, I have found reasons to contest the thesis defended by Causey (1977) and Schaffner (1993) according to which the reduction of a property leads to its identification with properties, or functions of properties, of the reductive theory. A property of a macroscopic object cannot, for logical reasons, be identical to properties of its microscopic components. The only properties with which it is logically possible to identify a systemic macroscopic property of a complex object are “structural” properties (Armstrong 1978) that Kim (1998) calls “micro-based properties.” A complex macroscopic object has such a micro-based property on the mere logical basis that it has a number of parts p₁, p₂, . . ., p_n, each with a number of properties P₁₁, P₁₂, . . ., P₂₁, . . ., P_nm, and that there are spatial relationships R among these parts. We have seen that the conditions for the existence of such a micro-based property are not sufficient to guarantee it a real existence in the sense of having causal powers of its own. Indeed, insofar as the existence of nomic interactions among the parts is not required, the mereological whole made up of my left shoe and your right shoe, or the mereological whole made up of the left hemisphere of my brain and the right hemisphere of your brain, have such micro-based properties. But in the absence of the relevant interactions, they have no real properties, in the sense of causal efficacy. No cognitive property emerges from the mereological whole made up of my left cerebral hemisphere and your right cerebral hemisphere. However, when the parts of such a mereological whole interact, it is possible that real systemic properties emerge that are qualitatively different from the properties of the parts. The laws of interaction between the atomic components of the hydrogen molecule give rise to the stable structure of the molecule. The laws of interaction that govern the interaction between the neurons and neuronal networks in my brain give rise to the thoughts that I am in the process of transcribing onto paper. If the interaction gives rise to a whole with emergent properties, then we can say that the emergent property is determined in a non-causal way by a “law of composition.”

My analysis has shown that the notions of reduction and “level of reality” are crucial for physicalism. These notions are indispensable for reconciling the thesis that the microphysical level determines all real states, processes, and events, with the reality of entities that are not microphysical, where “reality” means that these entities have causal powers of their own. Physical states of affairs determine other levels of reality but only by means of laws of composition that sometimes cannot be reduced to microscopic laws alone — as in the case of the thermodynamic hypothesis of equilibrium — and that generally cannot be derived a priori from knowledge of the microscopic level alone. We can understand the relationship among the different levels of reality only on the basis of knowledge acquired through observations and experiments conducted at all levels.

But the notion of levels of reality would not be fully justified if it were only a question of classifying phenomena as objects of knowledge. The thesis that levels of reality really exist, and are not just an effect of perspective generated by our fragmented approach to reality, can be justified only within the framework of a doctrine of emergence. Chemical phenomena have their own reality and their own causal powers in relation to microphysical causal powers only insofar as they are objectively different from microphysical phenomena. My analysis of the notion of emergence in Chapter 4 has enabled us to give this notion an ontological meaning that allows us to account for the qualitative difference between the phenomena that make up the different levels of reality. Interactions among objects belonging to a given level lead to the appearance or “emergence” of complex objects with properties qualitatively different from those of their components. These qualitatively new properties justify the idea that the complex objects that possess them form a level of reality that differs from that of their parts. Atoms occupy a relatively fundamental level of reality. The interactions among atoms described by quantum physics give rise to molecules and macroscopic objects, in particular solid bodies. Solid bodies possess many new properties that the atoms of which they are composed do not: solids are hard or malleable, transparent or opaque and coloured, whereas atoms cannot have any of these properties.

Living things occupy another level that seems to be clearly distinct in its characteristic properties. Living things organize and reproduce themselves, whereas their components do not. There can be controversy about the delimitation of the different levels characterized by specific properties. Here again the main way of overcoming these controversies is scientific: the existence of a level gives rise to a set of specific phenomena that are the subject of a specific science. The very existence of chemistry attests to the fact that there is a set of specifically chemical phenomena as well as a set of regularities that chemistry takes into account. Similarly, the existence of biology gives us reason to believe in the existence of a level specific to living beings and their properties. Finally, the existence of psychology gives us reason to believe in the existence of a level specific to cognitive systems and their properties. Controversies can arise from the difficulty of judging the status of different subdisciplines and their specific objects and properties: we can debate whether plants, fluids, and atmospheres of planets constitute separate levels, because they are the subject of botany, a subdiscipline of biology, and hydrodynamics and meteorology, subdisciplines of physics. Of course, it is also essential to accompany our judgment of the existence of a level with the prudent precaution that the history of science teaches us: subdisciplines and even main scientific disciplines appear and disappear. Thus, we have reason to believe in the existence of a psychological level of reality only since the birth of scientific psychology in the nineteenth century. The analysis of the concept of reduction also shows that the reduction of the properties and theories of one level to the properties and theories of lower levels often leads to the appearance of a new discipline that appears to be hybrid from the point of view of the disciplines existing before the reduction. The reduction of an elementary part of chemistry to physics gave rise to physical chemistry; the reduction of certain cognitive abilities to neuroscience gave rise to cognitive neuroscience. It is only by looking back in the long run that we can hope to make a well-founded judgment of the nature of the levels that correspond to these sciences.

Fortunately, it is not necessary for us to enter into these controversies and answer these difficult questions. I have achieved my goal if I have succeeded in making cognition and cognitive properties appear as occupants of a distinct level of reality, in the same way as chemical and biological objects and properties. It is from this general perspective that I have answered the question of whether cognition and its properties are “nothing other” than physical objects and properties and in what sense. Are cognitive phenomena mere physical phenomena conceived differently, with a different conceptual apparatus, as is claimed by the a priori implication thesis examined in Chapter 2 and defended for mental phenomena other than qualia, among others, by Chalmers (1996), Jackson (1998), and Kim (1998)? Are there mental phenomena that, conversely, are irreducible (both conceptually and empirically) to physical states of affairs and laws, as these authors maintain that qualia are? I have shown that negative answers to these two questions are at least consistent. It is conceivable that cognitive properties emerge from neurophysiological properties, so that they constitute their own level of reality with their own laws and causal powers, in the same way as chemical and biological properties. It is up to psychology, neuroscience, and especially the new science “between levels” — cognitive neuroscience — to show whether this conception really corresponds to the relationship between our minds and our brains.

We might consider construing levels of reality in terms of causal interactions: we might consider defining a level as the set of objects with which a given object, or a given kind of object, is capable of causal interaction. In a similar way, we might consider defining the characteristic properties of a level as the set of properties that enter into relationships of causal responsibility. Atoms interact primarily with atoms; macroscopic objects, such as our bodies, interact primarily with other macroscopic objects. However, this criterion does not lead to a clear delimitation of distinct levels because there are causal relationships between entities that intuitively belong to different levels. When a subject perceives, in a psychophysical experiment, an isolated photon absorbed by her retina, an elementary object, the photon, causes a cognitive effect situated at the level of the person. It therefore seems to be more sensible to ground the concept of level of reality on the existence of a set of properties that is the subject of a specific science and integrated into a set of laws of nature that is the object of the theories of that science.

To justify the intuition that cognitive properties are properties of persons (or animals) qualitatively distinct from the properties of their parts, we need to show that they are emergent. Therefore, it is important to find a criterion of emergence in the ontological sense. We have seen that physicalism imposes a certain number of necessary conditions for emergence. Emergent properties belong to objects composed entirely of physical parts, and they are determined exclusively by the physical properties of these parts and their interactions. The emergent properties of an object are systemic in the sense of not belonging to the parts of the object. However, these are only the necessary conditions for emergence that are also satisfied by properties not intuitively emergent: the property of a stone of weighing 5 kg does not seem to be emergent, even though none of its parts has it and even though it is determined exclusively by the properties of the parts of the stone.

I have suggested that there are mathematical criteria that can yield a sufficient condition for emergence at least for certain types of systems. For emergent physical properties, the topology of a system’s trajectory in a phase space can provide such a criterion: when a purely quantitative change in the properties of a system’s components can change the topological structure of the system’s trajectory, that trajectory is emergent. The scientific discovery of this topological structure can justify the judgment that the system has an emergent property. I have shown that the topological difference between the psychological space of representation and the physical space of the represented stimuli allows us to apply this topological criterion to certain emergent mental properties. It remains to be determined whether it is possible to generalize the application of this criterion to other cognitive properties. In the meantime, the existence of nomic regularities at the cognitive level, which intuitively seem to be qualitatively different from the regularities that characterize neuronal processes, gives us reason to believe that there is indeed a distinct cognitive level of reality.

My thesis that there are levels of reality with their own causal powers faces two objections, which I considered in Chapters 3 and 5. One influential view is that mental properties are dispositions. To have learned to associate a conditional stimulus (CS) with the unconditional stimulus (US) that triggers the response R is to have the disposition to react with R to the perception of the CS. The idea that mental states correspond to functional roles is the common heritage of analytical behaviourism and functionalism. The essence of a cognitive state consists of what causes it and what it causes independently of the intrinsic structure of the cognitive system. Ever since the polemic against the occult powers of medieval philosophy, likened to the “dormitive virtues” of opium, dispositions have been taken to be properties of dubious reality. It seems to be gratuitous to postulate the existence of a dormitive virtue in opium when it comes to identifying the property that causes the smoker to fall asleep. Instead of identifying, in a scientific manner, a real and intrinsic property of opium causally responsible for sleep, the postulate of a disposition to induce sleep seems to create only the illusion of knowledge. In Chapter 3, I considered a number of traditional objections to the reality of dispositions, which are also objections to their causal efficacy. It became clear that these objections do not refute my conception of causally efficacious properties, according to which they can be construed both dispositionally and categorically. Even if many mental properties are indeed conceived according to their functional roles, there is nothing to prevent the occupants of these roles from being macroscopic mental properties.

According to the functionalist theory of dispositions, the occupants of the roles that characterize dispositions are always microscopic categorical properties. In the same vein, the functional model of reduction developed by Kim (1998) for mental properties proposes that the causally efficacious properties that occupy mental roles are microstructural properties. According to this model, what makes the animal conditioned to the CS react by R are neuronal and biochemical properties or “micro-based” properties, which correspond to logical constructions from microscopic properties. However, if causal efficacy lies exclusively at the microphysical level, then it follows that there are no macroscopic properties and in particular no cognitive properties. The cognitive level is merely a conceptual level to which corresponds no level of real properties. The mental thus appears as epiphenomenal.

We have seen that it is possible to avoid this conclusion on the condition that a clear distinction is made between two meanings of “reduction” and between two meanings of “realization.” Showing that a categorical macroscopic property occupies a given functional role, or “realizes” the role, constitutes a first step that I have called “role-occupant reduction.” It is only at a second step that a macroscopic property is reduced, in the sense of microreduction, to microscopic properties or mechanisms. In one sense, the occupant realizes the role. Hemoglobin, for example, performs the role of oxygen carrier in mammalian blood. But we can also say, in another sense of “realize,” that the microproperties that determine a macroproperty in the non-causal sense realize it. The hemoglobin macromolecule is realized, in this second sense, by a certain chain of amino acids that are its microscopic components. Both forms of realization are compatible with multi-realizability. The function of transporting oxygen is multi-realized, in the first sense of realization, by hemoglobin, hemerythrin, and hemocyanin. Hemoglobin is a macromolecule multi-realized, in the second sense of realization, by different chains of amino acids.

It is conceivable that mental properties are emergent properties of cognitive systems that fulfill the roles defined by cognitive concepts. They can be causally efficacious even though they are determined by microscopic, neuronal, and biochemical properties and even though they can therefore be micro-reduced. In Chapter 5, I considered another important objection to the reality of mental properties: even if they are emergent and categorically conceivable, the principle of the causal closure of the physical domain and the principle of causal-explanatory exclusion seem to be incompatible with the idea that emergent properties exert their own causal influence on the course of physical events. According to the principle of causal closure, any physical event, at any previous moment, has a complete and exclusively physical cause. This principle is based on the observation that physics never discovers intrusions from non-physical causes. Therefore, at the moment when I make a decision, the physical consequences of that decision have an exclusively physical cause. If there is a complete physical cause at the moment of the decision, then what causal contribution could the decision itself make to a given physical consequence of my action caused by my decision? The principle of causal-explanatory exclusion states that, if causal “overdetermination” exists, then it is not systematic. Only in exceptional situations do two causal chains converge on the same event. This excludes the possibility that the physical causes of the consequences of our actions are systematically accompanied by parallel mental causes. From this reasoning, Kim (1998) draws the conclusion that there are no mental causes. Mental concepts are second-order concepts that quantify over first-order properties, which hold the monopoly of causal efficacy and, in his view, are always physical.

I have proposed the following way of avoiding the conclusion that there are no efficacious mental properties. I accept the principle of causal-explanatory exclusion. However, we have seen that physicalism does not oblige us to accept the principle of the causal closure of the physical domain. What is justified is a weaker principle according to which, at every instant prior to a given physical event, there exists a set of physical states of affairs that determines the event in question. Unlike the principle of causal closure, this determination can include a non-causal stage. Let us say that I decide to close the window, and this decision leads to the event of closing the window. There is a set of neural events underlying the decision that determines it in a non-causal way. It is conceivable that the decision contributes causally to the event of closing the window, along the lines of the following scenario. There is no microscopic law that determines at the neuronal level alone the detailed evolution of the complex system that is the person and her brain, in the sense that the evolution could be predicted or explained on the basis of the knowledge of the neuronal state with finite precision. However, the mental property of making this decision imposes a constraint on the system to evolve in the direction of an action that leads to closing the window. It is only a constraint because, first, many other mental properties exert influences on the evolution of the system, so that the regularities generated by each of these cognitive constraints are never strict, and because, second, the mental property alone does not determine the evolution of the system in microscopic detail; it determines its evolution only at the cognitive level. Each one of the stages in the preparation and execution of the action is compatible with many underlying physiological states. It is the preceding physiological state that determines which one of the physiological states compatible with a given cognitive state is actually realized. In this way, my closing the window, at the moment of my decision, has a complex cause: the decision as a mental property is indispensable to the causal determination of the action, and the underlying neurological state determines the evolution of the system among the possibilities compatible with the evolution imposed by psychological regularity. The principle of causal-explanatory closure is respected because there is only one complete explanation for the closing of the window. This explanation is partly mental and partly physiological. However, insofar as the determination of the events caused by our minds includes a non-physical aspect, the principle of causal closure is not respected. But physicalism does not require such a strong principle. My scenario is compatible with a principle of physical determination of all physical events: each physical event, at each instant preceding it, has a set of physical states of affairs that determines it entirely.

I end this conclusion with the remark with which I began it. It is up to the various relevant sciences, and above all neuroscience and psychology, to establish whether certain processes by which our minds seem to intervene in the physical world really correspond to the scenario outlined here. The justification of the reality and causal efficacy of mental properties belongs, at least in part, to science. The properly philosophical objective that I set for myself has been achieved if I have succeeded in showing that it is at least conceivable.