When a quantum physicist speaks about “charm,” she probably means not sexual appeal, but a subatomic elementary particle. Neurobiology, however, cannot do the same. If it claims to explain (or explain away) such things as consciousness, free will, or human language, it cannot give them its own, freely invented definition, but should start with the same meaning of these words that has already existed in general, philosophical, and scientific parlance before neurobiology starts its enterprise.

The target article (TA) clearly demonstrates that integrated information theory (IIT) fails as an attempt to explain consciousness. However, the TA is less clear regarding the question why IIT fails. From my viewpoint, the primary reason of the failure is the IIT's inability to define, on the first step, what it intends to explain. This inability is shared by many other neurophysiological approaches to consciousness that systematically disregard the definition of the explanandum.

Therefore, we should first agree about the necessary metacriteria for any neurophysiological or (more generally) biological theory of consciousness. These metacriteria should define (a) consciousness and (b) the doing word “to explain.” The latter is of high importance but has been discussed for centuries from the Antique (Evans, Reference Evans1959) to our days (e.g., Bechtel, Reference Bechtel1994; Keil, Reference Keil2006). To save space, I will only concentrate on the former.

Any theoretical approach claiming to explain consciousness or to find its reliable correlates should start with an explanandum that fulfills certain criteria that are (1) objective phenomenological, (2) empirical, (3) domain-specific, and (4) non-circular. As regards objectivity, it is a hard concept itself, but what I mean is at least its weakest notion, that is, intersubjective consensus. Although each of us may have a little different idea about what our consciousness is, there are nevertheless a few points which most people agree upon. Hardly, anybody would object that we have subjective experience (e.g., pain) and that many aspects of our consciousness possess the quality of “aboutness” (intentionality).

As regards empirical criteria, they are twofold. On the one hand, there is very rich literature (first summarized by Baars [Reference Baars1988] more than 30 years ago but having become yet richer later on) about strong effects of conscious awareness on experimentally observed behavioral outcomes. On the other hand, we urgently need social-psychological empirical studies about what lay people actually mean when they speak about “being aware of something.” Another debate on the border between philosophy and neuroscience, the debate about free will has been strongly contributed by such empirical studies demonstrating the real content of the beliefs in freedom and choice (e.g., Monroe & Malle, Reference Monroe and Malle2010; Vonasch & Baumeister, Reference Vonasch and Baumeister2013). Similarly, to attain a meaningful neurophysiological account of consciousness we must know how the corresponding terms are practically used by both experts and lay people.

As regards domain specificity, an explanatory theory can intend to explain a specific variety of consciousness, for example, full-blown human consciousness (Kotchoubey, Reference Kotchoubey2018), or much-discussed animal consciousness (e.g., Panksepp, Reference Panksepp2005; Seth, Baars, & Edelman, Reference Seth, Baars and Edelman2005), or that of severely damaged neurological patients just above the level of unresponsive wakefulness syndrome (Kotchoubey, Vogel, Lang, & Müller, Reference Kotchoubey, Vogel, Lang and Müller2014). In such cases, we require specific criteria characterizing exactly these varieties. Vice versa, a theory pretending to explain consciousness in general (like IIT) should be careful to avoid domain-specific features and to start with formulation of general criteria characterizing all kinds of consciousness.

Finally, the meta-criterion “non-circular,” meaning that the explanandum must not be described in terms deduced from the explaining theory, appears almost self-evident; but the inspection of the literature shows that the appearance is deceptive. There is a trend to confounding between features, correlates, and underlying mechanisms of consciousness. However, features are explanandum, (putative) mechanisms are explanans, and correlates are explanans when we are looking for correlates but can be a component of explanandum when we are looking for mechanisms. For instance, Seth et al. (Reference Seth, Baars and Edelman2005) include such “criteria of consciousness” as “Irregular, low-amplitude, and fast electrical activity,” dependence on thalamocortical loops, and “widespread (in contrast to local) brain activity.” But nobody experiences such things in their consciousness. I do not object that the criteria can be useful when working with individuals whose abilities to ordinary communication are lacking, for example, small children or animals. For example, the probability for a severe, behaviorally non-responsive patient to be still conscious is higher if his/her cortical activity is between 12 and 70 Hz than if it is between 3 and 7 Hz. But such diagnostic issues should be strictly separated from theoretical issues. We cannot search for neurophysiological correlates of consciousness whose definition includes its neurophysiological correlates.

Other examples of such circular criteria are “sensory binding” (Seth et al., Reference Seth, Baars and Edelman2005), synchronization of electroencephalogram (EEG) gamma oscillations, and “neural complexity” (Seth, Dienes, Cleeremans, Overgaard, & Pessoa, Reference Seth, Dienes, Cleeremans, Overgaard and Pessoa2008). Sensory binding, that is, the presumed ability of the conscious brain to build complex gestalt images from singular “features,” is a hypothetical process, a part of a theory (which has also been criticized by many, e.g., Palanka & DeAngelis, Reference Palanka and DeAngelis2005; Shadlen & Movshon, Reference Shadlen and Movshon1999). Even if the theory works, it is probably limited by visual modality, as gestalt mechanisms in audition may be completely different (Rauschecker, Reference Rauschecker1998), but we nevertheless experience auditory events consciously. Gamma synchronization is a strong correlate of conscious processes in some specific situations, but how far the correlation can be generalized and whether it is more than a correlation, remains unclear. Most probably, the synchronization is one of many necessary but not sufficient activation mechanisms underlying cognition (Merker, Reference Merker2013). The term “neural complexity” is obviously taken from IIT and is, therefore, an explicit example of circular reasoning.

We should be conscious of our own systematic flaw that, unfortunately, characterizes many approaches to the study of highly important aspects of human behavior such as freedom, aggressivity, or consciousness. Missing the clear definition of the explanandum, we slowly redefine it in such a way as to create something that we can hope to explain (Kotchoubey et al., Reference Kotchoubey, Tretter, Braun, Draguhn, Fuchs and Tschacher2016). But, after having explained this entity, we find that it may have nothing in common with the original meaning of the same concept.