Search

In their contribution to The Cognitive Neurosciences V, the most recent addition to the popular series of works of reference among cognitive (neuro-)scientists, Christiansen and Muller (2015) discuss the role of phylogenetically ancient neural circuits in the evolution of the human language faculty (FL) and its ontogenetic development from the perspective of what they refer to as “cultural recycling.” While I am in principle highly sympathetic to any endeavor that addresses “Darwin's problem” (Boeckx, 2009) within the conceptual framework of “neural recycling” (e.g., Dehaene and Cohen, 2007), it occurs to me that Christiansen and Muller's outright dismissal of Universal Grammar (UG) is premature and ill-founded, as they reiterate misconceptions regarding evolutionary biology and UG that are common in the literature (e.g., Dunbar, 2003; Christiansen and Chater, 2008; Christiansen et al., 2009; Evans and Levinson, 2009; Chater and Christiansen, 2010). Christiansen and Muller (2015) rightly call attention to the fact that the evolution of UG, that is seemingly (highly) arbitrary linguistic principles, by means of natural selection as, for example, put forward by Pinker and Bloom (1990) actually delineates a next to impossible scenario for human cognitive phylogeny. Clearly, the arbitrariness of UG principles would constitute a “moving target” for natural selection, that is a target that could hardly, if at all, have been selected for (Chater et al., 2009). The authors present this conclusion as evidence against the existence of UG and base their case for “cultural evolution” on it. Yet, in so doing they fail to fully explore the consequences of this claim. As Berwick (2009) already pointed out, an independent movement in theoretical linguistics has come to strikingly similar conclusions and has carried on where Christiansen and Muller (2015) seemingly chose to leave off. The assertion that FL did not “evolve” (in a neo-Darwinian sense) might appear somewhat bizarre at first, but the idea that natural selection is not the only force in evolution was already acknowledged by Darwin himself (in later editions of Origin of Species) and has been gaining widespread acceptance among biologists in the last two or so decades (Fodor and Piattelli-Palmarini, 2010/2011). Along these lines the so-called language as a “spandrel” (Gould and Lewontin, 1979) scenario has repeatedly and plausibly been put forward in the literature (e.g., Hauser et al., 2002; Fitch et al., 2005; Bolhuis et al., 2014). In essence, the argument is that FL, being universal among humans (Anderson and Lightfoot, 2000; Berwick et al., 2013; Hauser et al., 2014), arose “recently” and “suddenly” (Tattersall, 2009, 2013), a stance that deems large parts of FL to be exaptations (Fitch, 2010, 2011). Such an exaptationist train of thought seems to me to be, in principle, highly compatible with Christiansen and Muller's goal of painting a picture of FL within the conceptual framework of neural recycling during phylogeny. In the terminology of Hauser et al. (2002), the faculty of language in a broad sense (FLB) should be expected to have inherited properties of the evolutionary ancient systems of which it is composed so that the novelty of the narrow language faculty (FLN) might even come down to nothing else than the way in which it links these ancient systems together. Thus, the object of interest is the evolution of language as “[…] a computational cognitive mechanism that has hierarchical syntactic structure at its core […]” (Bolhuis et al., 2014, p. 1). Pressingly, we find that the design of this mechanism is often ill-suited for communicative purposes, instead adhering to principles of efficient computation (e.g., Chomsky, 2011). Because UG cannot have evolved through natural selection Christiansen and Muller introduce “cultural recycling” of preexisting neural structures as a supposed alternative. Despite the authors' statement that “Without our brains, there would be no language.” (Christiansen and Muller, 2015, p. 676), they nevertheless assert that as the brain did not adapt for language, language(s) instead must have adapted to human brains through “cultural evolution.” Given that there is no way for an I-language to “exist” apart from the human mind/brain (Chomsky, 1986), it is completely unclear how a language that did not “fit” the human brain could ever have existed (the hypothetical realm of the theorist's design space notwithstanding). Consequently, the “close fit” between language design and the human brain is, contrary to Christiansen and Muller's assessment, neither surprising nor in need of explanation. As a case in point, given the pace of language change (i.e., “cultural evolution”), all properties of language that counteract communicative efficiency (Christiansen and Muller's “socio-pragmatic considerations”) should long have been eliminated if there was no UG, provided that FL emerged roughly 100,000 years ago (Tattersall, 2009, 2013; Bolhuis et al., 2014). Yet contrary to Christiansen and Muller's conjecture this is not the case. The corresponding conclusion is that natural languages do not evolve in a (neo-)Darwinian sense, instead they change (Chomsky, 2011) within a rather narrow framework of possible variation provided by UG (currently best captured by the concept of parametric variation; see Baker, 2001). This fixed framework concerns linguistic features as opposed to their culturally determined values (Berwick, 2009), a distinction that the authors fail to make. Thus, accepting that the linguistic genotype is fixed (Anderson and Lightfoot, 2000; Berwick et al., 2013; Hauser et al., 2014), as it would seem that Christiansen and Muller (2015) do, language change becomes irrelevant for the study of the phylogeny of FL, and the role of “cultural evolution” is rendered even more obscure. Christiansen and Muller's (2015) dismissal of UG in part also rests upon a discussion of the multiple cognitive functions that have credibly been localized to Broca's area (i.e., Brodmann area 44/45), a canonical “language region.” But from an exaptationist point of view this is not surprising, after all, the evolutionary change that created FL is conceived as minimal so that we should expect recycling of the same neural tissue for different functions. Nevertheless, functional specificity for language has been demonstrated (Musso et al., 2003; Fedorenko et al., 2011), and the processing of hierarchical structure as a core linguistic operation has successfully been isolated from other functions attributed to Broca's area, such as working memory (Makuuchi et al., 2009). Furthermore, observed variability in the neural basis of the world's languages is generally highly restricted, whereas linguistic function (e.g., syntactic, lexico-semantic, etc.) wins out over form (i.e., encoding) in all known respects (Friederici and Ruschenmeyer, 2006; Friederici, 2011), even independent of modality (Emmorey, 2006). Unfortunately, available neuroimaging methods are not (yet) capable of capturing minute details with respect to (online) spatial and temporal resolution (Friederici and Ruschenmeyer, 2006; Kemmerer, 2015). However, everything currently known about the encoding of information in single neurons and cell assemblies (Quiroga et al., 2005; Sterling and Laughlin, 2015), as well as the fundamental principles on which much simpler nervous systems realize cognitive functions (Gallistel, 2009), points toward arbitrariness. While coding and computation are currently best understood in input-systems assigning “some order to our experience” (Anderson and Lightfoot, 2000, p. 15), such as the visual system famously studied by Hubel and Wiesel (1962), there is no reason to suppose that higher cognitive functions should be an exception. In general, the “atoms” of neural computation (Marcus et al., 2014) remain largely unknown and the granularity mismatch problem (Poeppel and Embick, 2005/2013) pertains, but efforts to ground linguists' abstractions in neurophysiology are under way (Friederici and Singer, 2015). What then to say in conclusion? The idea that UG might have arisen through the recombination of already existing neural systems with (highly) arbitrary features from which new (language-specific) ones could have emerged remains credible in the light of contemporary neuroscience. The basic premise of Christiansen and Muller's (2015) proposed solution for Darwin's problem is interesting and fully compatible with recent independent efforts of theoretical linguists, but as yet I do not see any reason to throw out the baby with the bathwater by either dismissing UG or intermingling the study of language change with the study of the evolution of FL.

In his article “Different conceptions of mental illness: consequences for the association with patients” Helmchen rightly cautions against any kind of dogmatism in psychiatry, regardless of whether it is a social, a psychological or a biological one. Instead, he favors the biopsychosocial model as a remedy for “the narrowing of conceptions that depict only partial aspects of mental illness” (Helmchen, 2013, p. 3). The main criticism of this model is traditionally that it “borders on anarchy” because one can emphasize the “bio” if one wishes, or the “psycho” […], or the “social.” There is “no rationale why one heads in one direction or the other” (Ghaemi, 2009, p. 3). Against this alleged arbitrariness and vagueness of the integrative model Helmchen recommends basing it “on scientifically proven concepts” (Helmchen, 2013, p. 4). Yet, it is not quite clear in which relation the three elements of the integrative model should stand and what its proposed grounding on scientifically proven concepts amounts to. I assume that the biopsychosocial model either has to be based on biological facts or else it will remain arbitrary. But if it is based on biological facts—even if not exclusively—it will probably be charged with “biologism” in just the same way as current accounts of biological psychiatry. It is certainly true that a dogmatic overemphasis of the physiological side of the disorder-coin is ill-advised and in all likelihood to the disadvantage of the patient. Whether biological theories of the mind are in fact utterly brain-focused is, however, a point of contention. Admittedly, there are indeed voices that urge the concept of mental illness to be replaced by an account of brain disease (Bickle, 2006; Akil et al., 2010; Holsboer, 2010; White et al., 2012). But the vast majority of biological psychiatrists does try to understand the patient's personal situation. Even a hardboiled reductionist cannot avoid asking the respective patient about what she “feels.” The reason is simply that it is up to now impossible to read off the brain whether someone feels depressed or not, whether she has delusions or not. The causes of a mental illness on the one hand and the symptoms on the other are to be found on different levels. This points to the distinction between “explaining” and “understanding” Jaspers is so often cited with and which even a biological psychiatrist cannot—and will not—ignore. In his General Psychopathology Jaspers explains: “The units of phenomenology (e.g., hallucinations, modes of perception, etc.) are explained by bodily events. Complex meaningful connections in their turn are considered as units (e.g., a manic syndrome plus all its contents can be regarded as the effect of a cerebral process or of some emotional trauma such as the death of an intimate.”) (Jaspers, 1963, p. 305, my italics). Biological psychiatry is sometimes regarded as nothing but an ideology (Berger, 2001; McLaren, 2010; at least implicitly also Cohen, 1993). On a closer look, things are not that simple. In actual fact, when Helmchen defends the integrative model as a “didactic tool” (Helmchen, 2013, p. 4), this comes quite near even to Bickle who declares within his “ruthless reductionism” that “[h]euristically, higher level investigations and explanations are essential to neuroscience's development” (Bickle, 2006, p. 428). Similarly, Insel and Quirion who demand “that mental disorders be understood and treated as brain disorders” emphasize “the need for a sophisticated understanding of interpersonal relationships along with the use of evidence-based, nonpharmacological treatments” (Insel and Quirion, 2005, p. 2223). Last but not least, Kandel—who, too, is usually regarded a radical reductionist—explains that it “would be unfortunate, even tragic, if the rich insights that have come from psychoanalysis were to be lost in the rapprochement between psychiatry and the biological sciences” (Kandel, 1998, p. 467). What is, then, the difference between the biological and the biopsychosocial model as Helmchen conceives it? According to the first, mental illness as a phenomenon can be understood on the level of the patient's experiences and be explained biologically. According to the latter, the psychological and social levels are essential for an adequate understanding of mental illness, but both should to be based on scientific proven concepts. The much criticized biological psychiatry seems not to be very different from the integrative, scientifically based model. A critic may object that the searched-for foundation of the biopsychosocial model of mental illness does not at all need to be a biological one. What about Psychology? Isn't it a science, too, with proven concepts? The answer to this objection is—in a nutshell—2-fold. Either psychology is a science or it is not. If it is, it is unavoidable to bring it into accordance with biology, which undoubtedly is a science as well. In cases of theoretical conflict we would need a criterion to decide which concept (or theory, or law) should be given priority. The crucial question, then, is whether the “bio,” the “psycho,” or the “social” should be the maestro in the orchestra of concepts. If, on the other hand, psychology is not a science, the psychologist will simply not be able to provide the “scientifically proven concepts” which are so necessary to mitigate the lurking arbitrariness of the biopsychosocial model of mental illness. It is one thing to use a conception as an auxiliary means for an adequate understanding of something and quite another to make use of it as a scientific foundation. To sum up, the biopsychosocial model of mental illness is valuable as a reminder that there is more to mental illness than brain functions. Seen as a theory, it will either be based on biology and meet similar trouble as the so called biologism in psychiatry, or else it will indeed be vague and border on anarchy.

Walter (2013) argues in that a bunch of externalist approaches examined in the theory of mind is interesting for psychiatry—especially for biological psychiatry. The externalist approaches state that mental states are not only constituted by internal but also by external factors. He subsumes them under the so-called 4E-thesis: embodied, embedded, extended, enacted. His example is concerned with ADHD such that ADHD is only existent “in an environment that favors attentional distraction and punishes hyperactivity.” First, using the 4E-thesis seems to be explanatorily attractive: There are factors of the disorder that can be grouped as external constituents and these constituents are not named by internalist vocabulary. For example, punishment is for sure not part of internalist vocabulary. But there is a figure of a principal argument against using the 4E-thesis in psychiatry. Consider this cookbook theory of the 4E-thesis: Cookbook theory for the 4E-thesis in accord with Henrik Walter's multilevel-approach: There is the phenomenon of a psychiatric disorder instanced in the form of multilevel-complexes. There are embodied, embedded, extended, or enacted (4E) constituents of cognition that realize the cognitive systems. There are individuation criteria for the 4E-constituents. Coupled systems of individuated 4E-constituents individuate a multilevel-complex. There are individuation criteria for systems of constituents C that realize multilevel-complexes. For example, with respect to extended cognition, the cookbook theory is the following: Cookbook theory of extended cognition: There is the phenomenon of cognition of a group of patients in psychiatry instanced in the form of cognitive systems of a specific structure. There are vehicles of cognition that realize the cognitive systems named in 1. There are individuation criteria for vehicles named in 2. Coupled systems of individuated vehicles individuate a cognitive system. There are individuation criteria for systems of vehicles of cognition that realize cognitive systems named in 1. As one can see, every attempt of the 4E-thesis needs criteria marking where to stop adding externalist constituents to the constituent base of the psychiatric disorder. For example, extended cognition-constituents should not include the whole world to make the thesis an interesting thesis—otherwise every phenomenon to be explained by extended cognition is constituted by the same constituents (i.e., the whole world). But what kind of criterion can the 4E-externalist provide us with? First, the 4E-externalist could give a criterion based on intrinsic properties of systems realizing disorders. But as a detailed analysis of criteria of this kind by Adams and Aizawa (2001, 2010) shows, these criteria are regularly based on finding a mark of the cognitive—which seems to be a hopeless endeavor without being a fundamentalist who just adds axiomatically this criterion to the cookbook theory. But second, the 4E-externalist could maybe give a criterion based on extrinsic properties of systems realizing psychiatric disorders. For example, one could use reliability as a stop-criterion for delineating what is coupled (with respect to point 4 in the cookbook theories). What is reliably coupled builds a constituent base of a system realizing a psychiatric disorder. But as a detailed analysis shows, this is not the case (Bukow and Will, in press). With respect to different criteria of reliability, there are very different reliable systems. This is not trivial, because in accord with the cookbook theory-approach, Bukow and Will argue: Argument for the arbitrariness of the reliability-predicate: Assume the cookbook theory of extended cognition. Give a definition of reliability with an operationalization (e.g., reproducibility as an operationalization). Choose an experimental procedure to test for reliability based on interaction between systems (because reliability is relative to this experimental procedure). Choose a value X%, i.e., that the experimental procedure is reproducible in X% of all series of experiments. There is no established decision procedure for the needed choices in points 3 and 4. And whatever procedure you choose, you may get different values for X% which will lead the psychiatrist to different constituent bases of the psychiatric disorder to be explained. Bukow and Will analyze in detail other candidates for criteria based on extrinsic properties, for example functional roles, causal roles, or heuristics. Now, there are two horns for the 4E-externalist: Horn 1: Use a stop-criterion based on intrinsic properties. The critics of Adam and Aizawa bite the proponent of this kind of criterion. If you add the criterion axiomatically, then you are a fundamentalist and not a scientist. Horn 2: Use a stop-criterion based on extrinsic properties. Then, there is the need to show in detail how to use such a criterion—which seems to be hopeless in accord with the investigation above. If you add the criterion without detailed analysis, then you are using just so-stories, without guarantee that the criterion fulfills its job. For these reasons, using the 4E-thesis is tempting but dangerous—there is no well-founded stop-criterion in philosophy of mind that a psychiatrist could use—not in a detailed way and not in a rough way. It is arbitrary what type and what token the psychiatrist may use with respect to the selection of constituents of a disorder. In a principal debate, the 4E-proponent may excuse these deficits because nobody may give in principal such stop-criteria. But in psychiatry, principal deficits are no excuse for misuse. Arbitrariness can smooth the way for use—but also for misuse in psychiatry, which is a dangerous zone with respect to rights and needs of patients of psychiatry. And if a psychiatrist can only give arbitrary constituent bases to a patient's disorder, then disorder and treatment will only arbitrarily be defined. This is, as far as I can see, an objection against Henrik Walter's argument for a mature “third wave” biological psychiatry.