It has been suggested that ethical and social norms are, from a perspective in which the brain is central, ultimately encoded as spatiotemporal patterns of neuronal activity that can be mobilized within the conscious neuronal workspace (Dehaene & Changeux 2011). Yet from a neurobiological standpoint, this view hinges upon the classical issue of the combinatorial explosion raised by the immense network of almost a million billion (1015) interconnected synapses of the human brain. The question that arises, then, is how the particular patterns of neuronal activity, which, for instance, encode defined actions or perceptual events and ultimately ethical rules, are selected within this gigantic neural network. In my view, the concept of a hierarchical organisation of the brain needs to be taken into consideration more closely.
Analysis of the neurological deficits caused by lesions discloses hierarchical and parallel neural architectures that help us understand higher brain functions (Shallice & Cooper 2011). Among these is the inhibition of automatic (or reflex) actions and the elaboration of goal-directed behaviours and their control. In the brain, an evolutionary-recent territory of cerebral cortex architecture, the lateral prefrontal cortex, has been shown to play a critical role in the temporal control of behaviour. It serves as a “temporal buffer” between past events and future actions, allowing behaviours that follow internal goals to occur (Fuster 2001; Goldman-Rakic 1987; Petrides 2005). Moreover, the lateral prefrontal cortex exerts top-down control of cognitive processes associated with hierarchically-lower regions distributed in more posterior territories on the basis of internal plans, goals, or what may be referred to as “rules” (Miller & Cohen 2001; Passingham 1993; Shallice 1988; Dehaene & Changeux 1991; Koechlin et al. 2003). It thus contributes to decision-making within the actual context of a given individual history and stored memories (Damasio 1994) and to “neurally encoded rules” that can associate a context with a specific behavioural response and the ability to generalize a rule in novel circumstances.
An early formal model of learning by selection according to a rule was devised in the Wisconsin Card Sorting Task, which is commonly used as a test of the integrity of frontal lobe functions (Dehaene & Changeux 1991). It requires subjects to infer a “rule” according to which a deck of cards must be sorted, i.e., colour, shape, or number. Feedback from the experimenter takes the form of a simple positive or negative reward (correct or incorrect). The goal for the subject is to get as many “right” responses as possible. Initially, cards must be sorted according to, say, colour. When performance is successful, the “sorting rule” is changed, for example from colour to shape; the subject must notice the change and find the new rule. The global architecture of a network that passes the task comprises two distinct levels of organization: a low level (level 1) that governs the orientation of the organism toward an object with a defined feature and which would correspond to a visuo-motor loop, including visual areas and the premotor cortex; and a high level (level 2) that controls the behavioural task according to a memory rule, and which would be homologous to the prefrontal cortex or closely-related areas (Dehaene et al. 1987; Dehaene & Changeux 1989).
A key feature of the model is that the high level contains a particular category or cluster of neurons, referred to as “rule-coding clusters”, each of which codes a single dimension (e.g., number, colour, or shape). During the acquisition step, the layer of rule-coding neurons is assumed to play the role of a “generator of diversity”. The spontaneous activity then plays a critical role in the activation of a given rule-coding cluster; and because of lateral inhibition only one cluster is active at a time. A search by trial and error takes place, until a positive reward is received from the environment (here the experimenter). Then, the particular cluster active at this precise moment is selected (for discussion see Monchi et al. 2001; Asplund et al. 2010; Fuster 2008). The number of trials necessary to learn the current rule is small (1–2), and single trial learning may occur in normal subjects as it does with the model (Dehaene & Changeux 1991). This learning of short-term rules based upon the fast (millisecond to second) allosteric transitions of synaptic receptors may also be transferred to long-term stores as epigenetically-acquired patterns of connections (see above).
In the course of the modelling of the Wisconsin card-sorting task, an additional architecture was introduced in the form of an auto-evaluation loop, which can short-circuit the reward input from the exterior. It allows for an internal evaluation of covert motor intentions without actualizing them as behaviours, but instead by testing them by comparison with memorized former experiences (Dehaene & Changeux 1991).
In these early formulations, the “rule-coding clusters” were pre-wired in the neuronal network. Subsequent models, however, opened the range of possible epigenetic rules to a brain-wide space of combinations made available within the global neuronal workspace (Baars 1988). This is of importance when we consider the ability to coordinate thoughts or actions in relation to internal goals, which is referred to as “cognitive control” and is a rather infrequent phenomenon. This discussion thus illustrates how rules encoding ethical norms may originate from the brain. Against this background—which shows how ethical rules might be epigenetically built from brain organization—I propose the possibility of being epigenetically proactive, and adapting our social structures, in both the short- and long-term, to benefit, influence, and constructively interact with the ever-developing neuronal architecture of our brains.
Cognitive control has been further investigated by Koechlin et al. (2003) using a set of more complex tasks than the Wisconsin Card Sorting Task, and which span (at least three) nested levels of complexity. They consist in the presentation of series’ of coloured visual stimuli (squares or letters) organized into blocks, with an increasing importance of contextual signals: from “sensory control” with little if any contextual signal, to “contextual control” and, at the higher level, to “episodic control”. Brain imaging fMRI recordings with healthy human subjects revealed that the lateral prefrontal cortex contributes to a hierarchical cascade of executive processes that involve at least three nested levels of processing. These are neurally implemented in distinct regions, from posterior premotor to rostral lateral prefrontal cortex regions (typically Brodman’s area 46; Koechlin et al. 2003; Badre & D’Esposito 2007; Badre et al. 2009). Patients with focal lateral prefrontal cortex lesions performed cognitive tasks with sensory, contextual, and episodic deficits associated with focal damage to Brodman’s areas 6, 45, and 46, respectively—as is expected from the cascade model (Azuar et al. 2014; Kayser & D’Esposito 2013).
By analogy with the Wisconsin Card Sorting Task (WCST) model mentioned above, behavioural rules are also sorted, but at different nested levels of information processing, the highest level rules “controlling” in a top-down manner the underlying rules closer to the senses. Hypothetically, ethical norms may be viewed as some particular kind of “control rules” developed within a social context, though this possibility still deserves to be explored by Koechlin, D’Esposito and colleagues.
Recently Collins & Koechlin (2012) have further suggested a computational model of human executive functioning associated with the prefrontal cortex, which integrates multiple processes during decision-making, such as expectedness of uncertainty, task switching, and reinforcement learning. The model reveals that the human frontal function may monitor up to three or four concurrent behavioural strategies and infers online their ability to predict action outcomes: whenever one appears more reliable than unreliable, this strategy is chosen to guide the selection and learning of actions that maximize rewards (see also Miller & Cohen 2001; Passingham 1993; Shallice 1988; Fuster 2008; Dehaene & Changeux 2011).
In their original paper, Collins and Koechlin do not explicitly mention social interaction. Yet we may consider an extension of their model to the social context by assuming that ethical or social norms are part of the “concurrent behavioural strategies” that they postulate exist in decision-making. The selection and learning of actions would then be more elaborate than the simple maximization of immediate rewards.
The developing baby is exposed very early on to a defined social and cultural environment, possibly even pre-natally (Lagercrantz & Changeux 2009; Lagercrantz et al. 2010). At this stage of development an intense synaptogenesis steadily occurs in the cerebral cortex, and epigenetic selection of neuronal networks accompanies the acquisition of the “maternal” language as well as of the common rules of the social community to which the child’s family belongs. The developing baby/child is “impregnated” with the current ethical rules of the social community, and this is often linked with the symbolic (philosophical/religious) system of representation character of the community to which it belongs. These early traces may last for the lifetime of the individual and sooner or later create conflicting relationships with a fast-evolving environment aggravated by the increased longevity of the individual (Changeux 1985). On the basis of the neurobiological data mentioned above, one may define these rules as epigenetically-acquired patterns of connections (scaffoldings) stored in frontal cortex long-term memory, which frame the genesis of novel representation and “cognitively controlled” decision-making in a top-down manner.
Against this background I propose the possibility of being epigenetically proactive and adapting our social structures, in both the short- and the long-term, to benefit, influence, and constructively interact with the ever-developing neuronal architecture of our brains.