Anti-Representationalism and Mirror-Neurons-Based Action Recognition Models: Ideomotor

Massimiliano Cappuccio

Mirror neurons and skilful coping:

motor intentionality between sensorimotor and ideo-motor schemata

ingoal-directed actions

to appear in

Ricardo Pietrobon, The Research on Research, Sussex Press, UK 2009 (forthcoming)

Abstract

I approach philosophicallymirror neuron theory by situating itbetween the sensorimotor and the ideo-motor approaches to skilful coping. Mirror neuron theorists havedeveloped their account of goal-oriented action under the influence of Merleau-Ponty’s concept of motor intentionality. I compare it to the Merleau-pontian description of motor intentionality proposedby Hubert Dreyfus, showing that his account stresses prominently a sensorimotor model of skilful coping, and that – for this reason – it can only partly matchthe embodiment of motor goals in mirror functions. I analyze then Wolfgang Prinz’s model of William James’ motor ideas, showing that it is symmetrical under many aspects to Dreyfus’ model, and that – for this reason – it explains only the other half of the motor intentional process. I propose to consider sensorimotor and ideo-motor approaches to goal-oriented actions as reciprocally linked and complementary. From the phenomenological point of view, this means that, while motor intentionality instantiated by the former aims to an optimal equilibrium between body and world, the motor intentionality instantiated by the latter aims to break this equilibrium. From the empirical point of view, it means that the former derives motor goals through a bottom-up process of inductive abstraction of kinaesthetic invariants, while the latter filters the meaning of kinaesthetic experiences by means of a top-down process which is shaped within the boundaries of primary goal-related motor ideas.I show that mirror functions can constitute the condition of reversibility for the two approaches; but I also stress that only mirror functions codify the intentional meaning of the action in afully-embodied way, because they express goals in anoriginary motor format.

Mirror neurons and the definition of goal-oriented intentional actions

Experiments carried out by Giacomo Rizzolatti’s parmesan research group show that mirror-neurons-related motor schemata of purposeful transitive behavioursare activated not only when a subject is executing an action, but also when he is observing the same action while it is performed. Recordings with microelectrodes implanted in single cells of macaque ventral premotor cortex show that each one of the different mirror neuron circuits responds selectively to a type of transitive goal-oriented actions like grasping, placing, holding, crushing, ripping, bringing to mouth[1].That’s why mirror neuron theory affirms that motor schemata related to the activation of mirror neurons pertainto goal-oriented actions and express, at once, an executive valence and a cognitive-perceptual one. Differential analysis[2] has progressively made clear that their response does depend neither on the identity of the agent (potentially anyone performing the action)[3], nor on the object targeted by the action (as long as it allows that specific kind of action), nor on the involved effectors (left hand, right hand, mouth or tools, as long as they allow the same kind of practical intervention), nor on the kinematics of the gesture (e.g. grasping from the top, from the bottom[4], or through anon-natural kinematical strategy by means of a tool[5]),nor on distality(and this excludes the influence of a volitional attitude, i.e. the desireto performthe gesture).

The criteria for mirror neuronactivation are tolerant enough to result insensitive to different kinematical strategies and to different effectors involved, as long as they aim to the same goal; at the same time, mirror neuron circuits are selective enough to distinguish extremely similar kinematical strategies, as long as they display different goals.This means that the same “grasping” mirror neuron circuit in the macaque brains is activated by very heterogeneous movementswith hands, mouth or pliers, as long as they are meant to grasp something, but the same movements do not trigger the activation of mirror neurons if they are not directed to that concrete goal. From the point of view of the embodied motor skills of a macaque, it is only secondary what trajectories are drawn by its movements and what body parts are involved in order to grasp a nut with the fingers, but it is primarily relevant that the macaque grasps itwith an expectation of a successful result.

By undermining the syntactical role of the kinematical structure,the experiments havealways confirmed that every mirror neuron circuitcorresponds specifically to the intentional, purposeful meaning of the kind of action it is associated to (Fogassi et al. 2005), independently from the fact that the action isactuallyperformed or merely perceived:in the former case we shall argue that the intentional meaning is actually expressed by the body as an executive behavioural schema accessed in a first-person-perspective, while in the latter it seems that the intentional meaning is accessed in a third-person-perspective and recognized asa potentially executable purposeful motor schema. The intentional meaning is accessible then in either a performative or in a perceptual way, but what is most important is thatit is always mapped in a motor format, accessibleduringexecution and recognition tasks. The discovery of mirror neurons is extremely relevant just because they consist in premotor structures that display cognitive functions besides control functions: this suggests that the agent, by means of his own bodily proficiencies, already govern a “motor knowledge”[6] of the potentially disposable actionsthat is pre-reflectively operating before and below the higher levels of intelligent processes.The classical cognitivistic paradigm of motricity is put into question by this discovery, because it shows that motor and premotor areas are not only involved in execution and control of the action, but also in cognitive tasks related to recognition, categorization and prediction[7] of the intentional meaning of the actions.

This new frameworkgives finally flesh to a fully embodied scheme of cognition, because the knowledge of basic goals of motor actions is intrinsic to the executive/perceptual processes and not derived from them: sensesand motricityare not the two opposite extremities of the main process of elaboration (like in the classical sandwich-like cognitivistic framework: “input-computation-output”).The motor process is in factsintrinsically endowed with a cognitive meaning which is quasi-independentfrom previous or higher processes of elaboration. In a few words, this confirms a classical principle of Merleau-pontian phenomenology according to whichthebodyalready holdssome knowledge about how toact intentionallybeforeand below any mental process of representational or symbolicalelaboration concerning howthe actionis structured[8]. This principle resonates with Rizzolatti’s idea that mirror neuron circuits form a built-in vocabulary of simple intentional goals expressed in motor terms[9], codified as open-ended schemata for general practical intentions, and not as instantiations of single definite movements.From the evolutionary, the functionalist and the phenomenological point of view, the basic motor repertoires are more efficiently and economically organized, retrieved and enacted by patterns of expected motor effects (goals) than by patterns of movements.

It is phenomenologically relevant that, when an agent performs a purposeful action, the conscious experience of acting in a goal-oriented waydependsneither on the complex of stimuli that he perceives from his body, noron the topological coordinates of his anatomical parts; and also when a subject recognizes the intentional meaning of an action performed by someone else, the recognitionis derived by the holistic evidence of the meaningfulpurposeof the action, and not by some neutral elaboration of the meaninglesstopological modifications occurring in the agent’s system of body parts[10]. This phenomenological evidence finds in mirror neuron theory a promising counterpart for what concerns its sub-agential neurological implementation. Given the priority of the holistic purposeful meaning over the collection of disconnected sensorimotor stimuli, the discovery of mirror neurons confirms this phenomenological evidence by suggesting that proprioception and somatosensory information are not the only required elements (and not the most fundamental ones) in order to control/recognize goal-oriented intentional actions[11].

Experiments on mirror neurons have shown that the intentional meaning concerns purposes as it corresponds to the “how-to” knowledge that associates motor goals to bodily actions. Mirror functions are functionally defined neither by the syntaxof the kinematic structure of simple movements(the coordinates that have been altered in the topology of the body-environment system), nor by thesemantics of the concepts behind the motivation of the action (beliefs, knowledge, mental representations and so on); rather, they are definedby the pragmatics of the action(i.e. the consequences of the action for the life of the agent and for its essential embodied experience).This pragmatist principle circumscribes the applicability of the cognitivistic assumption according to which the goal of the action is a formal content (a mental representation) ontologically independent from the material conditions in which the body operates and logically separable from the environmental situation in which the action takes place[12]. The intentional meaning of the action is not a formal semantic structure exportable and identically reproducible in infinite contexts, because the bodily action and its environmental scenario – in order to be meaningful – must belong to each other and can never be completely isolated. No dualism between body and world can account for the intentionality embedded in mirror-neurons-related practical schemes of actions, because the motor-goal is embodied in the very movements allowing the action, and the action itself is embedded with the contextual meaning of the environment in which it takes place: in facts, the purposeful schemata of the intentional actions can’t be defined independently from their bodily realization (even if multiple movements can instantiate the same motor purpose), and motor schemata are always associated with the open-ended range of their pertinent environmental contexts (even if different contexts can solicit the same motor response)[13]. But, if the goal of the action is not derived from the perceived movements and is noteven a concept, what is it then?

Attractor basins and mirror neurons:motor intentionality and skilful coping according to Hubert Dreyfus

Phenomenology helps to answer this question describingthe dynamic systemic relationship undergoing between goal, body and environment in intentional actionsrelatedto mirror neurons. This relationshipis accounted for bythe notion of motor intentionality, that Giacomo Rizzolatti and Corrado Sinigaglia (2007b) have developedby re-interpreting acentral notionof Maurice Merleau-Ponty’s Phenomenology of Perception. During the last years, motor intentionality has also been in the spotlight of an important debate in philosophy of mind, which flourished around Hubert Dreyfus’ decisive article “A Merleau-Pontyian Critique of Husserl's and Searle's Representationalist Accounts of Action” (2000), and around Sean Dorrance Kelly’s studies (2000, 2002) on the Phenomenology of Perception.In the present context it seems particularly interesting to compare the interpretations of motor intentionality given by mirror neuron theorists and by Dreyfus. According to Merleau-Ponty’s phenomenology, and to the influentialphilosophical interpretation that Dreyfus has given of it, motor intentionality consists in a sort of tensionexperiencedthrough the body; this tension isendowed with purposeful meaning andmotivatesgoal-oriented actions like the ones enacted by mirror neurons. According to this phenomenological point of view, motor intentionality is inseparable from its bodily context and from the environment in which action takes place, and this happens to be necessarily truefor a very simple reason: according to a sensorimotor account of action, an agent performs an intentional motor-actionas soon as he is seeking an adequate response of his body to the environmental situation; consequently,the intentional meaning of the action is exactly determined by the ongoingdynamic relation of adequacy/inadequacy between body and world.

According to Dreyfus, the refinement of already familiarmotor skills,or the acquisition of new ones, is mediated by the agent’s capability of modulating bodily responses to solicitations offered by the situations; this capability doesn’t consist in a simply mechanical reactivity or in a set of instructions stored in the subject’s mind, but in “a disposition” of the body to compensate thecontextual circumstances. The development of these “skilful” dispositions is governed by the agent’s “sense of an optimal gestalt”: the agent’s body gradually modifies its responses to the perceived solicitations in accordance with environmental feedbacks. This process goes on until the agent’s bodycomes to fit perfectly the environment, reaching a dynamical equilibrium with it. The action of the agent, modulated by the body-environment situation, is clearly intelligent and intentional, because it aims to the fulfillment of a practical purpose; nonetheless, during this process, the agent doesn’t access any representation of the optimal gestalt, nor he possesses an already defined set of behavioural rules that would consent him to plan his actions; the gestalt toward which his actions tend is not mentally depicted but it is simply sensed as aforce camp attracting the agent’s bodily movements. The agent doesn’t foreseein advance the situation in which his body will be when it will have accomplished its motor purpose, but he can sense how distant from the optimal situation his body is, and he can sense in which direction he should orient his movements in order to reduce the sense of inadequacy engendered by this distance.

The process necessary toachieve theoptimal state is called“maximal grip”by Merleau-Ponty, and indicates that an intentional agent is motivated to adapt to the situation in the smoothest and most harmoniousway through his body, so to get closer and closer to the condition of virtual equilibrium that Dreyfus has called“optimal [or satisfactory]gestalt”. The continuous adaptive relation between agent and world is called “intentional arc”: this concept presupposes thatintentionally structured reactive behaviors (coping) are endowed with the meaning of their conscious (intentional) embodied experience; the phenomenological notion of intentional arccontains the idea that the meaning of the action and its contextual background are not only strongly associated, but alsoco-determinated bytheir relation of reciprocity, so that every environment displays anintentional meaningby means of the actions that might be performed in it, and every action has someintentional meaning because of its possibility to entertainrelations with endless environmental situations. This kind ofgoal-oriented actions(not only the actually performed ones, but also the potentially disposable ones) modulates the relation between the agent’s body and its environmental situation, so that both of them owe their intentional meaning from the agent’s capability of acting adequately.

Dreyfus has acknowledged that not every intentional motor action is dependent onsuch intelligent, goal-oriented, highly adaptive, pre-reflective, not-necessarily aware andnon-representational process, since many intelligent behaviors, especially in human beings, are explicitly planned and rationally structured[14]. Dreyfus’ account is restricted to coping, as it consists ina very fundamental bodily activity that is prominently guided and controlled by motor intentionality. From now on, I’m going to use the term “skilful coping”, already used by Dreyfus in his exchange with Michael Wheeler[15],in order to indicatethe intentional actions that instantiate specifically goal-directedprocesses of harmonization between body and world. Each kind of skilful coping is endowed with a distinct finalistic sense that, however, depends only from the bodilycapacity to counterbalance the environmental stimuliso to reduce the sense of deviation from a satisfactory gestalt. According to the mathematical models of artificial neural nets developed by neuroscientist Walter Freeman[16], and revisited by Dreyfus in a phenomenological frame, each purposeful action describable as skilful coping is guided by its specific attractor basin, which governs neuronal activity by bringing it to reach in a specific way its minimal level of energy; each attractor basin is phenomenologically equivalent to the gestaltic force camp that guides and reshapesthe motor experience by soliciting the agent toachieve a satisfactory adaptation to its contextual situation.

It is very interesting that Dreyfus-Freeman’s account of skilful coping seems at leasthypoteticallycompatible– under some conditions - with mirror neuron theory, and notably with the theory of a vocabulary of basic motor goals.Mirror neurons theorists explicitly mention the Merleau-pontian concept of motor intentionality for describing the dualperformative/perceptual competence that is possessed by the body and that is enacted by the premotor cortex; this confirms that Merleau-Ponty’s phenomenology of embodiment is still offeringpriceless suggestions to neuroscientific research and to philosophical studies. Apparently, an important part of mirror neurons theory can be immediately described in the framework provided by Dreyfusian analysis: if we think that each kind of simple, adaptive and goal-oriented actions corresponding to a mirror neuron circuitconsists in a specific type of skilful coping, we are easily brought to assume that this circuit might work like (or together with) an attractor basin which directs the motor intentionality of the agent toward the fulfilment of his bodily purpose, without that any representation of the purposewere necessary. This would account for mirror neurons capacity of codifying actions which are endowed with motor goals, which are simple, non-representationally codified, although theirkinematics is structured in an open-ended way that is alsohighly adaptive to different practical contexts.

Both mirror neurons and skilful-coping-related attractor basins are concerned with specific goal-oriented intentional skills possessed by the body, and this suggests that the two theories can be reciprocally enriched and clarified by their unification, as long as this is allowed by their own specific assumptions. In reality this unification would be possible only ifthe functioning of the attractor basins for motor intentionalitymight displaya basilar performative valence and a secondary perceptual one at the same time, exactly like mirror systems do; two different types of experiential contents (perception and motor activity) should be completely matched (1), and moreover their information should be stored in a motor format (2).

About (1), it is plausible that Freeman’s attractor basins could matchperceptual states (e.g. a particular smell of food, like banana)and motor reactions (biting the banana)[17]. Motor reactions can produce a “reward” for the perceptual activity and thus strengthen the functional disposition of the basins. The basinsmight map in a twinned, temporary and elastic structure both perceptual stimuli and motor responses, but they are still two functionally separate moments of the cognitive process; also mirror neurons match perceptual stimuli and motor programs, but their matching is deeply rooted in pragmatics, and it is also more strict and specific, since the perceptual stimuli that are matched with the motor responses pertain exactly to the very intentional meaning of theaction that is codified by the circuit (then the action related to that circuitis not coupled with whatever sensory feedback, but with the specific perceptual aspectsof the goal-oriented action itself).In both cases, anyway, the perceptual elicitation of the neural structures initialize a behavioural program that not necessarily has to be expressed, in facts it can just be inhibited and made present as a simple virtual opportunity of intervention in the world; but it is crucial that this potential consists just in a perceptual schema according to Freeman’s basins, while in the case of mirror neurons it displays a concrete practicalmotorsignificance, concerning how the action must be structured.