Enactive sense of agency

A new open access paper discussing the enactive notion of sensorimotor agency and how it helps explain the phenomenology of the sense of agency in a non-representational manner.

Buhrmann, T., and Di Paolo, E. (2015) The sense of agency – a phenomenological consequence of enacting sensorimotor schemes, Phenomenology and the Cognitive Sciences, doi: 10.1007/s11097-015-9446-7. (online first).

Abstract. The sensorimotor approach to perception addresses various aspects of perceptual experience, but not the subjectivity of intentional action. Conversely, the problem that current accounts of the sense of agency deal with is primarily one of subjectivity. But the proposed models, based on internal signal comparisons, arguably fail to make the transition from subpersonal computations to personal experience. In this paper we suggest an alternative direction towards explaining the sense of agency by braiding three theoretical strands: a world-involving, dynamical interpretation of the sensorimotor approach, an enactive description of sensorimotor agency as contrasted with organic agency in general, and a dynamical theory of equilibration within and between sensorimotor schemes. On this new account, the sense of oneself as the author of one’s own actions corresponds to what we experience during the ongoing adventure of establishing, losing, and re-establishing meaningful interactions with the world. The meaningful relation between agent and world is given by the precarious constitution of sensorimotor agency as a self-asserting network of schemes and dispositions. Acts are owned as they adaptively assert the constitution of the agent. Thus, awareness for different aspects of agency experience, such as the initiation of action, the effort exerted in controlling it, or the achievement of the desired effect, can be accounted for by processes involved in maintaining the sensorimotor organization that enables these interactions with the world. We discuss these processes in detail from a non-representational, dynamical perspective and show how they cohere with the personal experience of agency.

Keywords Enactive cognitive science. Agency. Sense of agency. Sensorimotor contingencies. Equilibration. Metastability

Advertisements

Non-representational sensorimotor knowledge

Close your eyes and follow the contour of the table in front of you. When you reach the end of it, do you think you would be able to continue moving your hand along the same imaginary line as if the table went on a little further? Most people would not have any problem doing this for a wide variety of conditions (body posture, distance and angle with respect to the table, etc.) Given this variety of circumstances, traditional approaches to perception and motor control postulate when we’re no longer in contact with the table, we must be using some form of representation that keeps guiding the arm along the same invisible line in a robust manner.

Model-eps-converted

Together with Thomas Buhrmann, we have demonstrated that representations are not necessarily the only explanation, and that indeed it is possible to attune the movement of your arm to the world without at any point involving any internal models or representations, but simply by a shaping of dynamical sensorimotor transients. The “memory” of the direction of movement is kept in the whole agent-environment system in a way that generalizes across various relative positions and angles of movement.

Manifolds2b-eps-converted

For more information see:

Buhrmann, T. and Di Paolo, E. A. (2014). Non-representational sensorimotor knowledge. In A.P. del Pobil et al. (Eds.): From Animals to Animats 13, Proceedings of the 13th International Conference on Simulation of Adaptive Behavior, SAB 2014, LNAI 8575, pp. 21–31, NY, Springer Verlag.

Piaget, dynamical systems, and how do we learn to perceive something new

Following our recent work on formalizing some aspects of the sensorimotor approach to perception – in particular the notion of sensorimotor contingencies – using dynamical systems theory, we have continued along similar lines and tackled the question of how we ever learn to perceive something new.

The sensorimotor approach proposes that perception is constituted by the possession (and depending on the interpretation, also the enactment) of skilful mastery of sensorimotor regularities, so that in order to perceive something as soft, one must already know how and where one should apply pressure to it. The question that emerges is how do we ever learn to perceive something new if, according to this approach, perception requires already having a sort of sensorimotor mastery. In this way, that which we know how to perceive is perceivable, but that which we don’t know how to perceive could never be perceived because why would we develop new forms of mastery in order to perceive something that we don’t even know it’s there?

The problem has a solution once we admit that the notions of mastery and skill can have degrees. At a subpersonal level it is possible then to build a theory that establishes how we progressively form new sensorimotor capabilities as much as a response to internal norms as through the guidance of the world. An example of such a theory is Piaget’s theory of equilibration.

fnhum-08-00551-g001Our recent paper:

Di Paolo EA, Barandiaran XE, Beaton M and Buhrmann T (2014) Learning to perceive in the sensorimotor approach: Piaget’s theory of equilibration interpreted dynamically. Front. Hum. Neurosci. 8:551. doi: 10.3389/fnhum.2014.00551

offers a reformulation of Piaget’s theory in the language of dynamical systems theory, which allows us to establish a strict compatibility with our previous formalization of sensorimotor contingencies.

We these developments it is now possible to start thinking of more operational notions of mastery, thus contributing to various debates in embodied cognitive science. We see mastery in non-representational terms, as a progression of dynamical relations of equilibration with the world (including the world of others, although this is not directly addressed in this paper).

Tarkovsky’s Stalker and the paradoxes of non-linear decision making

In Andrei Tarkovsky’s 1979 film Stalker three men undertake a perilous trip into the Zone, an off-limits and run-down post-industrial area full of the dangerous but valuable scattered debris left behind by a passing alien visitation (the leftovers of a “roadside picnic” as the title of the original novel by the Strugatsky brothers suggests). It is said that within the Zone there is a room with the power of making one’s innermost desires come true, even the unconscious ones – a ruthlessly rule-governed spot in a land of exception. The road to this room is full of mortal traps, often invisible to the eye. Your body could be crushed with artificial gravity or burned to cinders almost without a warning.

Two of the men, who for different motives want to visit the room, hire the help of a third as a guide, a Stalker. The guide is overcautious as they advance, making the others walk in line and for short stretches at a time. He tests for the presence of traps by throwing metal nuts with long ribbons into the air looking for distortions in their trajectory. At one point they come very close to the building where the room is located and yet the Stalker convinces the other men that they should not enter it directly but they should instead take a very long-winded route leading away from it. The journey is arduous and confusing and the travellers often find themselves stuck confronting the same situations and dilemmas. The route parallels the inner personal journeys of the characters and their own stumbling blocks.

Image

I believe that such a situation is an accurate, if extreme, description of how motivations, rational judgements and emotional reactions interact during ongoing decision-making in real life. It certainly provides a better picture than the way these questions have been investigated in neuroscience, psychology and economics until recently. Our decisions are rarely isolated events, and their interaction is rarely additive. And yet these are the basic assumptions of most research on decision-making and on the interplay between the rational and emotional processes that underlie it. Judging a situation and making a decision is generally embedded within a dynamical context in which a path towards a longer-term goal is being laid down by our own actions.

Consider a doctor prescribing a certain long treatment for a patient. The initial steps of the treatment may still serve a diagnostic function and as a result of one intervention, the doctor decides what the next step should be. Setbacks may as much indicate an insufficiency or an excess provoked by the treatment. And some interventions are deemed too risky to be used at first even though perhaps they would produce the desired effect and most quickly cure the patient. Past decisions, risk judgments and actions interact with not fully known time-extended processes in the world. Real life decision making is a non-linear journey into the unknown, where the unknown is constantly changing in part as a result of our own decisions.

In a recent study, Manuel Bedia from the University of Zaragoza and I have examined this situation using a non-linear model of Antonio Damasio’s Somatic Markers Hypothesis. Our mathematical model considers the case of a journey towards a goal similar to the situation depicted in Stalker in a much more simplified setting. According to Damasio’s hypothesis, and to  other so-called dual-process theories of decision making, the act of judging a situation and deciding on a course of action is not a simple cost-benefit analysis but a context-dependent mixture of rational and affective processes. The question we ask is: what counts as a good balance in this mixture? What factors determine how much we should rely on affective or rational processes?

Our findings were surprising at first. The answer to our question turns out to be highly dependent on whether we consider decisions in a chain as independent events (the traditional linear approach) or allow for interactive coupling between decisions, moves, and situations in the world.

We have found that the effect of primary emotions (often associated with conservative, cautious and protective actions) tend to be larger in non-linear decision chains than one would expect from their effect on single decision events. This results in over-cautiousness and leads to frequent re-visits to stumbling blocks until they are finally overcome. On average we tend to “stay on the spot” in a situation that again re-elicits the same emotional reactions that lead back to it in the first place.

Secondary emotions (somatic associations between past experiences and visceral processes) are conceived as making sense only if their positive effect is larger than the combination of rational and primary emotional decision making. But again, this is true only for an isolated decision event. It turns out that badly attuned somatic markers – for instance, reckless and unreliable gut feelings that often would lead us to bad results (e.g., losing our cool, replying to an annoying email too angrily and too quickly) – can sometimes break the paralysing effect of over-represented cautionary emotions. In other word, in non-linear decision chains, a combination of “bad” decision-making mechanisms (over-cautiousness and recklessness) can lead to a positive chance of successfully attaining your goal. This is what we have shown mathematically.

We might call this approach to decision making “enactive” in that decisions and actions affect the dynamics of the path that needs to be traveled towards a goal, a “path laid down in walking”. This famous phrase by poet Antonio Machado was picked by Francisco Varela and his followers to describe one of the central insights of enactive science: the fact that cognitive agents are participants and not merely observers. Apart from highlighting the importance of this perspective and showing that it makes a concrete, empirically testable difference, the result also has interesting developmental and evolutionary implications. It could help explain how somatic markers can exist in the first place if their positive effect depends on experiential tuning to the world and this tuning can only be poor at the early stages of development. If our result holds in such situations, it would indicate that in fact this is not a problem at all because even poorly attuned somatic markers can have a positive effect.

Childish recklessness is after all one way of breaking out of the paralysis of excessive precaution and learning about the world and making your mark in it. How else are you going to make your way through the Zone and reach that room?

Bedia M and Di Paolo EA (2012). Unreliable gut feelings can lead to correct decisions: The somatic marker hypothesis in non-linear decision chains. Front. Psychology 3:384. doi: 10.3389/fpsyg.2012.00384