It is a fact and an undisputable truth that there cannot be any mental life without the brain. More controversial is whether the level of description offered by the brain is also sufficient for providing a thorough and biologically-plausible account of social cognition. We think it isn’t. We would like to ground this assertion on two arguments: the first deals with the often overlooked intrinsic limitations of the approach adopting the brain level of description, particularly when the brain is considered in isolation and its intimate relation with the body is neglected; the second deals with social cognitive neuroscience’s current prevalent explanatory objectives and contents.
The contemporary emphasis divulgated by the popular media, namely the supposedly revolutionary heuristic value of cognitive neuroscience, mostly rests upon the results of brain imaging techniques, and in particular on fMRI. fMRI is often presented as the ultimate method of investigation of the human mind. It should be pointed out, though, that fMRI studies do not constitute the whole story in cognitive neuroscience. Cognitive neuroscience can indeed carry out its investigation at a more drastic sub-personal level, such as at the level of single neurons (see below), both in macaque monkeys and, although much more rarely, even in humans. These alternative approaches notwithstanding, the main thrust of cognitive neuroscience in studying human brain function is, as we speak, mostly confined to fMRI.[7]
Unfortunately fMRI only indirectly “sees” the workings of the brain, by measuring neurons’ oxygen consumption. Such a measure is also indirect, as it depends on the local difference between oxygenated and deoxygenated hemoglobin—the iron-rich molecule housed by red blood cells, which carriess oxygen to all bodily organs and tissues. Oxygenated and deoxygenated hemoglobin have different paramagnetic behaviours in relation to the strong magnetic field that is created by a big coil, inside which the head is placed. The measure of this functional parameter allows scientists to estimate local neural activity in terms of different MRI (magnetic resonance imaging) signals. The indirect quality of this kind of estimation of brain activity, which is based on local hemodynamic brain responses, inevitably introduces distortions and noise. Indeed, when studying any sensorimotor, perceptual, or cognitive function, in order to maximize the so-called “signal-to-noise ratio”, several repetitions of the same task in many individuals are required.
This means that fMRI allows us to indirectly assess the average brain-activation level induced by any given task across a population of no less than twelve to fifteen different individuals. Within each studied individual brain the spatial resolution of fMRI is within the order of few millimetres. This implies that we are able to measure at best the potentially coherent activation pattern of several hundred thousand neighbouring neurons, possibly also differing among one another in terms of their excitatory or inhibitory role.
Temporal resolution is even worse, since it is in the order of a few seconds. One should consider that action potentials, or “spikes” as neurophysiologists like to call them—the electric code employed by neurons to “communicate” with each other, and ultimately the true essence of neurons’ activity—last less than one millisecond. fMRI cannot match such temporal resolution because it measures the delayed (of about two seconds) and prolonged (for about five seconds) local hemodynamic response providing neurons with all the oxygen their electric activity requires.
As we have previously argued, fMRI is not the only available experimental methodology for studying the brain. Many different techniques are available nowadays (e.g., PET (positron emission tomography), NIRS (near-infrared spectroscopy), Tdcs (transcranial direct-current stimulation) or TMS (transcranial magnetic stimulation)). Particularly, since the revolutionary introduction in 1927 by the Nobel Prize laureate Edgar Adrian (Adrian & Matthews 1927a, 1927b) of the extracellular microelectrode, which allows the recording of action potentials discharged by single neurons, neurophysiology has made enormous progress in revealing the brain’s physiological mechanisms. Such neurophysiological investigation started with the study of the neural circuits that preside over elementary sensorimotor behaviours, like spinal reflexes, finally moving all the way up to the investigation of action and perception, reward and emotions, spatial mapping and navigation, working memory, decision-making, etc., in behaving animals like macaque monkeys. Unfortunately, such a finely grained level of description—both in terms of spatial and temporal resolution, is most of the time precluded in humans.
We posit that the scientific study of intersubjectivity and the human self requires a comparative approach, and that this is the only one capable of connecting the distinctive traits of human nature to their likely phylogenetic precursors. In so doing, and by making use of the single-neuron recording approach, neurophysiological mechanisms and the cortico-cortical networks expressing them can be related with several aspects of primates’ social cognitive behaviour and thus be thoroughly investigated. Conceptual notions like intersubjectivity and the self should be analyzed in order to better understand their nature, structure, and properties. Such an analysis, which provides us a deflationary notion of the same concepts, intended as the identification of their minimal component and the detailed study of their origin, will be most successful if driven by a meticulous investigation of the underpinning neurophysiological mechanisms—which most of the time are available only from the study of non human primates’ brains, hence the necessity of a comparative perspective.
Human brain imaging, because of the intrinsic limitations we briefly outlined above, can only provide correlations between particular brain patterns of activation and particular behaviours or mental states. This implies that the correlation between a particular brain state and a particular phenomenal mental state of a given individual human being[8] is most informative when the specificity and uniqueness of such a correlation can be firmly established. Unfortunately, this is not always the case with fMRI studies. Very telling is the supposed mindreading specificity of some cortical circuits comprising the ventral portion of the mesial frontal cortex and the TPJ (temporo-parietal junction; e.g., Leslie 2005; Saxe 2006). Such specificity is not only so far unproven, but is actually confuted by accumulated evidence (for a lengthier discussion of this point and for arguments and experimental evidence against such specificity see Ammaniti & Gallese 2014; Gallese 2014).
In spite of all these limitations, this neuroimaging approach turned out to be very productive, enabling us to study for the first time in parallel brains, behaviour and cognition, shedding new light not only on human brain structure, but also on its wiring pattern of connectivity and many of its functions. If we put the newly-acquired knowledge on brain function provided by cognitive neuroscience under scrutiny, we can make very interesting discoveries. For example, we discovered that in many areas of investigation brain imaging replicates and validates at a different scale what had been previously discovered at the single neuron level in animals like macaques.[9]
The prominent discoveries, among others, of David Hubel, Torsten Wiesel, and Semir Zeki on the functional organization of primates’ cortical visual system, like the orientation-, shape-, motion- and colour-selectivity of visual neurons, were made by correlating the discharge activity of single neurons in macaques’ visual cortices with different parameters of the visual stimuli macaques were looking at (for a comprehensive review of this literature, see Zeki 1993). These results later promoted a similar investigation carried out on the human brain by means of fMRI. Remarkably enough, a similar functional architecture was detected in the human visual brain, in spite of the species difference and, most importantly, the different scale at which these investigations were carried out: a few hundreds recorded neurons at best, in the case of macaques’ brains, versus hundreds of thousands if not millions of activated neurons detected by a local increase in blood flow in the case of the human brain.
Face-selective neurons, first described in the early nineteen-seventies by Charles Gross and colleagues in macaques’ temporal cortex (Gross et al. 1972), and immediately ridiculed as “grand-mother cells”, offer another very telling example.[10] Face-selective brain circuits appear to be strikingly similar in macaques and humans (see Ku et al. 2011). Furthermore, even in human brains single neurons were detected that selectively respond to a single face—such as the so-called Jennifer Aniston’s selective neurons (see Quiroga et al. 2005). These neurons respond to multiple representations of a particular individual, regardless of the specific visual features of the picture used. Indeed, these neurons respond similarly to different pictures of the same person and even to his or her written or spoken name. The authors of this study claimed that their evidence supported the notion that single neurons within the human medial temporal lobe cortex instantiate the abstract representation of the identity of a single individual.
Such examples seem to suggest that in spite of the big scale magnification implied when confronting single-neuron data from macaques and fMRI results from humans, some important functional features are nevertheless manifest across these different levels of description (i.e., single neurons vs. brain areas). One can study canonical or mirror neurons (see below) by recording the activity of a few hundred spiking neurons from a behaving macaque monkey during object and action observation, respectively. The same results can be replicated by detecting, by means of fMRI, and during object and action observation, the simultaneous activation of hundreds of thousands of human neurons within analogous cortical areas of the human brain.
This remarkable but often neglected fact cannot be the result of a pure coincidence. This evidence should thus invite us to resist and argue against those who downplay the heuristic power of single-neuron recording. Their thesis is that because of a supposedly incommensurable gap between single neurons and the incredible complexity of the human brain, where information would be exclusively mapped at the level of large populations of poorly selective neurons, it doesn’t make any sense to study the brain by recording single neurons. The fact however is that in spite of the almost astronomic figures characterizing the human brain (about 100 billion neurons, each of which connects with thousands of other neurons), its complexity does not parallel such astronomic figures, or at least not in such a way as to deny any heuristic value to the single-neuron recording approach. Let’s see why.
As argued by Chittka & Niven (2009), brain size may have less of a relationship with behavioural repertoire and cognitive capacity than generally assumed. According to the same authors, larger brains are, in part at least, a consequence of larger neurons that are necessary in large animals due to basic biophysical constraints. Larger brains also contain greater replication of neuronal circuits, adding precision to sensory processes, detail to perception, more parallel processing, enlarged storage capacity, and greater plasticity. These advantages, maintain Chittka & Niven (2009), are unlikely to produce the qualitative shifts in behaviour that are often assumed to accompany increased brain size, or at least not in a one-to-one manner.
The evidence so far briefly reviewed and that we will present in the next sections suggest that some functional properties of the brain exhibit a sort of “fractal quality”, such that they can be appreciated at different scales and levels of investigation. For these reasons fMRI cannot be the sole neurocognitive approach to human social cognition,[11] but it must be complemented by other approaches compensating for some of its deficiencies: like TMS, EEG (electroencephalography), and the comparative functional study of non-human primates by means of brain imaging and single-neuron recordings.
After having clarified what we take to be the often-neglected limitations of cognitive neuroscience that may hinder its potential heuristic power (namely, that fMRI offers only an indirect estimation of brain activity, inferred by measuring neurons’ oxygen consumption, and this inevitably leads to distortions and noise, and that it does not provide good temporal resolution because it measures the delayed and prolonged hemodynamic responses due to neurons oxygenation), let us now move to the second argument against the sufficiency satisfaction condition of the current approach of cognitive neuroscience to the study of human social cognition. This argument concerns the explanatory goals and contents of contemporary mainstream cognitive neuroscience. Vast quarters within cognitive neuroscience are still today strongly influenced by classical cognitivism, on one side, and by evolutionary psychology on the other. Classical cognitive science is the bearer of a solipsistic vision of the mind, according to which focusing on the mind of the single individual is all that is required in order to define what a mind is and how it works. The image of the mind that classical cognitive science gives us is that of a functional system whose processes are described in terms of manipulations of informational symbols in accordance with a series of formal syntactic rules.
According to evolutionary psychology, by contrast, the human mind is a set of cognitive modules, each of which has been selected during evolution for its adaptive value. Major figures of this current, such as John Tooby and Leda Cosmides, have gone as far as maintaining that the brain is a physical system that works like a computer (Cosmides & Tooby 1997). According to Steven Pinker (1994, 1997), our cognitive life can be referred to in terms of the function of a series of modules like the linguistic module, the module for the Theory of the Mind, etc.
Based on this theoretical framework, in the last twenty years cognitive neuroscience when investigating human social cognition has mainly tried to locate—as mentioned above—the cognitive modules in the human brain. Such an approach suffers from ontological reductionism, because it reifies human subjectivity and intersubjectivity within a mass of neurons variously distributed in the brain. This ontological reductionism chooses as a level of description the activation of segregated cerebral areas or, at best, the activation of circuits that connect different areas and regions of the brain. However, if brain imaging is not backed up by a detailed phenomenological analysis of the perceptual, motor, and cognitive processes that it aims to study and—even more importantly—if the results are not interpreted, as previously argued, on the basis of the study of the activity of single neurons in animal models, and the study of clinical patients, then cognitive neuroscience, when exclusively consisting in brain imaging, loses much of its heuristic power. Without the demonstration of the specific correlation between brain states and mind states and the explanation of such correlation, much of the contemporary brain imaging approach to social cognition looks like a sort of high-tech version of phrenology.
For this reason a “phenomenologization” of cognitive neuroscience is desirable, as Gallese has proposed before (see Gallese 2007, 2009, 2011, 2014). In Gallese’s view, to “phenomenologize” cognitive neuroscience means to start neuroscientific research from the analysis of subjective experience and of the role that the living body plays in the constitution of our experience of material objects and of other living individuals. In this way, the empirical study of the genetic aspects of subjectivity and intersubjectivity can be pursued on new bases—if compared to those thus far adopted by classical cognitivism. Francisco Varela a few years ago realized a similar possibility and set out on a pathway of analysis in this direction (Varela & Shear 1999).
Times change, however. We are insisting on nothing less than a change of paradigm. A new neuroscientific approach to the study of the human mind is gaining momentum. It capitalizes upon the study of the bodily dimension of knowledge: the so-called “embodied cognition” approach. In the next section we introduce mirror neurons and embodied simulation. Our purpose is to show that, starting from a sub-personal neuroscientific description of the pragmatic relationship with the world, a pathway can be traced to define the forms of subjectivity and intersubjectivity that distinguish human nature, rooted in the bodily interacting nature of human beings.