Agency from a radical embodied

perspective:
An ecological-enactive approach
Miguel Segundo Ortin
ENSO Seminar Series
7th May 2020
“The relation between the schools of thought is one of strange
familiarity, as if their respective practitioners were staring at
each other across an uncanny valley” (Di Paolo et al., 2017, p.
18)
In the last years, many authors have tried to articulate a complementary relationship between ecological
psychology and enactivism: McGann (2014), Baggs & Chemero (2018), Heras-Escribano (2019),
Rietveld & Kiverstein (2018), etc.
Where I want to arrive…
I aim to explore whether we can build a radical embodied theory of
agency by combining enactivism and ecological psychology

Roughly, agency is here understood as the capacity of an individual

organism or system to execute goal-directed or intentional actions.

So conceived, agency is manifested whenever an individual “acts on its own

behalf in an environment” (Kauffman, 2000, p. 8) or when she does
“something by itself according to certain goals” (Barandiaran, 2009, p.
369).
Thesis

Enactivism and ecological psychology can complement each other to

explain agency.

While enactivism focuses on investigating how the history of interactions of

an organism gives rise to a series of sensorimotor schemes and habits that in
turn play a causal role in determining how she interacts with the affordances
of the environment, ecological psychology, through the notion of ecological
information, explains how the individual can access to the environment’s
affordances without mediating representations and inferences.
The main inspiration(s)

A division of explanatory tasks between ecological psychology and

enactivism (Heras-Escribano, 2019)
But I disagree that the division maps onto the personal-subpersonal
distinction

Radical embodiment in two directions (Baggs & Chemero, 2018)

But I disagree that enactivism alone suffices to explain agency
Overview of the talk
• Agency at the ecological side

• Agency at the enactive side

• Towards a productive synthesis of both (to explain agency)

• Bridging the uncanny valley

Agency at the ecological
side
The three (main) principles of ecological psychology (Chemero, 2009):

(1) Perception is direct (non-mediated, non-inferential).

(2) Perception is active

(3) Perception is action-oriented (of affordances)

Perception as depicted by ecological psychologists presupposes agency
(see Brancazio & Segundo-Ortin, 2020):

• Perceptual information does not come to the animal. The animal actively seeks
for it. Therefore, perception is goal-oriented (J. J. Gibson, 1974[1982], p. 387-
388; Mace, 2015, p. x).
• Affordances are opportunities for action, not causes of action. It means that
perception of affordances is not enough to control behavior. The organism must
act upon these affordances. (Gibson, 1975[1982], p. 411)
• Because a single object offers multiple affordances to an animal, how she
behaves is unconstrained by the affordances of the object. The animal must
select what affordance to actualize at each moment. (Cutting, 1982)
Explaining agency is an old problem for ecological psychologists:

“The goal of ecological psychology is to explain agency scientifically, not

to explain it away or simply offer a discourse about it” (Reed, 1996, p. 19)

“To a large extent, scientific psychology is the science of agency – the

ability to select, perceive, and actualize affordances appropriately
based on intention” (Wagman, 2019, p. 148)
Reed’s approach (1993, 1996)
Reed characterizes agency as the capacity to put attention and action at the serve of intention (see also E. J.
Gibson & Rader, 1979):

“In any stuation, an individual’s intentions serve to select a small number of the potential affordances
available in that situation. This selection is reflected in the organization of the individual’s attention and
activity” (1993, p. 46)

He nonetheless disagrees that intentions are mental states that cause actions. Instead, he proposes:

“From an ecological point of view, intentions are not causes of action, but patterns of organization of
action’ they are not mental as opposed to physical, but are instead embodied in the kinds of performances
most likely found in cognitively capable creatures. … The development of an intention is thus the
development of the ability to nest bouts of exploratory and performatory behavior so as to achieve desired
outcomes” (p. 62)
How do intentions emerge (if not from the mind of the perceiver)?

• An intention can only emerge whenever there are multiple affordances available
for the organism.
• Reed takes inspiration from Darwinian evolutionary biology  Intentions, just
like any other biological entity, emerge out of processes of variation and selection.
• The minimal units of analysis are the Perception-Action Cycles (PACs). Each PAC
is specific to an individual affordance. Then, in situations where the organism is
offered multiple affordances, the PACs enter a sort of competition, and this
competition results in a goal-oriented or intentional behavior: the actualization of
an affordance: “Intentions are thus the ‘species’ that emerge out of competition
among perceptual and action processes for utilizing afforndaces” (1993, p. 65)

But, what are the mechanisms/selection pressures that define this competition?
Two options:

(1) Affordances themselves exert selection pressures: “the relative availability (or non-
availability) of affordances create selection pressures on the behavior of the individual
organism” (1996, p. 18)
Problems:
(a) Reed’s selectionist approach is purely passive/adaptationist. This goes against
current evolutionary biology and O-E Mutuality.
(b) We can think of actions that are either irrelevant (e.g., nail-biting) or harmful
(e.g., smoking).

(2) Socio-cultural norms influence competition between affordances (1993, 1996).

Problem: A theory of agency based solely on social norms cannot play the bill,
for there are non-human organisms capable of agency and that do not have social
norms.
Affordances as invitations (Withagen et al.,
2012, 2017)
Withagen et al. rely on contemporary phenomenology (especially Dreyfus).

They reject the notion of intention advanced by Reed. Instead, they claim that if
we want to understand agency in ecological terms, we must conveive of the
possibility that affordances can invite behavior:

“When actively exploring the environment, the agent is attracted and repelled
by some of its affordances, and the ensuing behavior is partly the result of
these invitations. This means that to understand how animals make their way in
the world, the inviting character of affordances should be taken central” (2012, p.
257)
They propose a list of features that are likely to be important for affordances to invite:
• Action capabilities of an agent and the amount of effort it takes to utilize an
affordance
• Evolution
• Cultural factors (see also Rietveld & Kiverstein, 2014)
• Individual’s history of interactions

Problem 1: This approach presupposes agency: “an affordance can invite behavior if and
only if an agent perceives it. If affordances are not perceived (or even have not been
discovered) they do not have the potential to attract (or repel) the according behavior of
the agent. Hence, a prerequisite for affordances to invite is an actually present
observer that actively explores the affordances of its environment” (2012, p. 257)
Problem 2: Invitations are not causes either, so we still have to explain “the animal’s
capacity to modulate the coupling strength with these affordances” (2017, p. 14;
Bruineberg & Rietveld, 2014 use FEP)
Baggs and Chemero (2018) propose that ecological psychologists should
adopt the theory of agency of enactivists:

“Combining the two would seem to provide a complete picture of

cognition: an enactive theory of agency, and an ecological story of the
environment to which the agent is coupled” (p. 2)
Agency at the enactive side
A characterization of agency
Agency entails three necessary and jointly sufficient conditions (Barandiaran et al., 2009)

• Individuality
• Interactional asymmetry
• Normativity

Di Paolo, E., Barandiaran, X, & Buhrmann, T. (2017).

Sensorimotor Life (fig. 5.5, p. 125)

We can find these conditions at different levels: biological, sensorimotor, social, and linguistic, etc.
Biological agency
Autonomy (organizational closure) is a necessary and sufficient
conditions for individuality. The most basic form of autonomy can be
seen in autopoietic systems

Autopoietic/autonomous systems need to interact with their

environments (if only to take nutrients and expulse waste products).
They must develop their own perspective of the environment 
“Sense-making”

Being a sense-maker implies being ready to selectively act upon the

affordances of the environment that are relevant to maintain
autonomy. For the sense-maker “the environment becomes a place of
valence, of attraction and repulsion, approach or escape” (Thompson,
2007, p. 158)  Sense-making is essential to understand agency
Autopoiesis provides a very thin understanding of biological agency.
Di Paolo (2005) proposes to combine autopoiesis with adaptivity:

Adaptive systems are a subclass of autopoietic systems that

implement second-order (meta-metabolic) processes which function
is to put environmental encounters in relation to the whole spectrum
of its viable states. This leads to recognizing the tendencies that lead
to the loss/improvement of autopoiesis.

Adaptive systems can recognize as meaningful environmental

features in virtue of their virtual consequences, thus perceiving
“graded differences between otherwise equally viable states” (p. 437)

Investigating the species-specific mechanisms that give rise to this

“graded perception” is a crucial to understand how different
organisms asymmetrically regulate their action in environments with
multiple competing affordances
Problem: The norms that arise out from biological
individuation concern to our survival only. Therefore,
if we were biological agents only:

(1) Anything that it is not relevant for our survival

would be irrelevant for us

(2) We would necessarily avoid any action that could

entail a risk to our biological integrity (e.g.,
smoking)
Sensorimotor agency

SA is manifested by sophisticated organisms capable of learning and

acquiring new behavioral repertoires  SA depends on BA, but it is
underdetermined by it.

The idea that justifies the extension of agency beyond the biological realm
is that we can find an autonomous entity at the level of perception and
action.
Sensorimotor schemes  Organized, mutually
adjusted, sequences of sensorimotor coordination
patterns the individual deploys in carrying out a
specific task, and that have been established as
preferable in light of some normative framework (e.g.,
considerations of efficacy, timing, precision, and so on)
SM Schemes can turn into habits: “self-sustaining
precarious SM schemes” (Di Paolo et al., 2017, p. 144)

Di Paolo, E., Barandiaran, X, & Buhrmann, T. (2017).

Sensorimotor Life (fig. 4.1, p. 85)
When a particular scheme is executed it has an effect on other
schemes, either preventing them from occurring or increasing
the likeliness they will be enacted.

SM schemes, just like SM patterns, can be organized in

“clusters,” this is, networks of mutually coherent and enabling
sensorimotor schemes  SM Networks

SM Schemes and Networks are regarded as the autonomous

individuals that give rise to sensorimotor agency:

“the behavioural analogue to biological agency is a network of

precarious but interactively self-sustaining sensorimotor schemes,
i.e. a self-asserting sensorimotor repertoire, whose adaptive
regulation is directed at the preservation of internal coherence
and consistency” (Buhrmann & Di Paolo, 2017, p. 219)
Di Paolo, E., Barandiaran, X, & Buhrmann, T. (2017).
Sensorimotor Life (fig. 6.4, p. 162)
SM Schemes and Networks define a set of viable
actions that contribute to their preservation, thus
conditioning how we relate to the external world.

They “become a source of [non-metabolic]

normativity for an agent, in such a way that the
preservation of her habitual identities guides much
of her perception, thoughts, and behaviors”
(Ramirez-Vizcaya & Froese, 2019, p. 7)

When considered at the sensorimotor level, the

affordances become relevant not only because they
contribute to our survival, but because they bestow
“the stability and coherence of [our] sensorimotor
repertoire” (Di Paolo et al., 2017, p. 39) Di Paolo, E., Barandiaran, X, & Buhrmann, T. (2017).
Sensorimotor Life (fig. 6.4, p. 162)
Studying sensorimotor agency requires investigating how sensorimotor schemes
develop and relate to each other forming complex identities that in turn affect the
way we interact with the affordances of the environment:

“It is literally a case of explaining who you are by referring to what you do, and
explaining what you do but referring to who you are” (Di Paolo et al., 2017, p.
142)
Towards a productive
synthesis
An “ecological plot twist”
SM schemes (and networks of these) are grounded in complex dynamical
arrangements of properties within the agent (e.g., muscle-skeletal structures, neural
networks, etc.) that in turn give rise to specific perception-action dispositions.

However, SM schemes are not something a body possess, but “modes in which
structures in the agent and structures in the environment meet and mutually stabilize”
(Di Paolo et al. 2017, p. 152). Therefore, SM schemes “constitutively involve both
the organismic body and the environment” (ibid.)

But, if the environment is constitutive of SM schemes, then agency is not a property

of the organism, but “a property of a relation between that system as its
surroundings” (p. 110)
If follows that to explain SM agency, this is, in order to account for how SM
schemes unfold, selectively exploiting affordances of the environment, we have to
account for how the organism can access the environmental structures that
complement their sensitivities and dispositions:

“Having powers and sensitivities required for action, in other words, is only half
of the story. The other half is access to suitable accompanying conditions
surrounding the agent” (Di Paolo, 2019, p. 212)

Without this complementary story, the enactive account of agency is

incomplete.
My proposal

Ecological psychology can provide enactivists with this

complementary story

What enables the organism to access the environmental structures that

are relevant to its goals is the existence of perceptual information of the
kind ecological psychology describes
According to EP, Perception consist of the un-mediated and non-inferential
detection of information.
Information is contingent on the existence of spatial-temporally extended and
structured patterns in the topology of the organism’s sensory array – “invariants”
Representational arguments in perceptual psychology depend on the assumption that
perceptual information is inherently ambiguous (or impoverished)
However, ecological information is grounded in lawful relationships:
• The presence of a particular property or event.α in the environment lawfully
generates a particular pattern β in the energy array that corresponds to (or
specifies) it  β guarantees α
• Then, β carries non-ambiguous information about α  Being aware of β
suffices to be aware of α.
“Direct perception can be possible if
properties of the world are specified
Lawfully generates
Property or in patterns of stimulus energy. If
event α in Invariant patterns of the world are
the pattern β unambiguously specified,
environment Guarantees the presence of perception does not have to involve
processes of interpreting
ambiguous cues” (Fajen et al., 2008,
p. 81)
Because information is
unambiguous about the properties
of the environment (or the O-E
system), then we can control our
action by relying on this
information.

“Affordances inform the individual

about its ongoing relation with the
environment, which permits the
selection of a course of action and
also the prospective control of the Turvey, M. & Carello, C. (1986). “The Ecological Approach to Perceiving-Acting. A pictorial
chosen mode of action” (Esteves et Essay”. Acta Psychologica 63: 133-155.

al., 2011, p. 640)

Therefore, rather than simply taking the enactive theory of sensorimotor agency, I hold that
ecological psychologists, through the notion of ecological perceptual information, can
contribute to explaining it.

A dual approach to agency:

• Enactivists focus primarily on describing how our acquired sensorimotor schemes and habits
mutually equilibrate, affecting our tendency to act upon some affordances instead of others,
• Ecological psychologists focus on studying how perceptual information contributes to the
actualization of sensorimotor habits without mediating representations, inferences, and
computations.

Agency is a property of the relation between the organism and its environment, where
this coupling is made possible by the existence of specific (unambiguous) task-related
perceptual information
Besides, ecological psychologists can contribute to enactivism in the following
ways:

(1) EP provides well-tested theoretical and empirical methods that allow for the
identification of the informational variables that need to be detected to enact a
particular scheme, carrying out its associated task (e.g., time-to-contact)

(2) EP provides a new characterization of sensorimotor mastery  Sensorimotor

mastery depends on embodying habitual SM schemes that integrate acton-
perception patters with task-specific perceptual information (e.g. dynamic
touch, motion parallax)
Bridging the uncanny valley
1. In EP, the activity of the perceiver plays no explanatory role

“For Gibson, these optical invariances, as well as the environmental properties they
specify, do not depend in any way upon the perceptually guided activity of the animal …
whereas Gibson claims that the environment is independent, we claim that it is enacted
… Thus the resulting research strategies are also fundamentally different: Gibsonians
treat perception in largely optical (albeit ecological) terms and so attempt to build
up the theory of perception almost entirely from the environment. Our approach,
however, proceeds by specifying the sensorimotor patterns that enable action to be
perceptually guided, and so we build up the theory of perception from the structural
coupling of the animal” (Varela et al., 1991[2016], p. 204)

“We agree with ecological psychologists when they highlight that real environments are
rich enough to access directly their relevant meaningful aspects. We think they are in
fact too rich, and that sense-making always involves a massive reduction of all the
environmental energies that might affect the agent” (Di Paolo et al., 2017, p. 227)
But:
(1) The affordances of the environment are perceiver-relative  They depend on body-scale properties, but also
on the perceiver’s action capabilities
(2) J. J. Gibson (1979[2015], p. 4) distinguishes between the world and the (animal-relative) environment 
Information is a property of the environment, not the world (Michaels & Carello, 1981; Segundo-Ortin et al.,
2019)
(3) Perception is an active process (something that the animal does)  It requires that the animal actively forages
for information by (i) focusing attention and (ii)moving in order to give rise to the required informational
patters (sometimes).

So, it is not true that EP denies the role of the perceiver in bringing about its own perceptual world.
2. EP vindicates functionalism and information-processing

“It is true that both approaches overlap in their rejection of

representationalism, but this does not mean they are necessarily rejecting the
same thing. … the enactive perspective rejects a functionalist general
approach to cognition, whereas ecological psychology rejects the
assumption of the poverty of environmental information… For
Gibsonians, perception is still about information pickup” (Di Paolo et al.,
2017, p. 18, ff. 3)

For Hutto & Myin (2017), the use of concepts such as “pick up” and “use’ of
information “suggests an underlying commitment to an information
processing story that is certainly inconsistent with [the radical embodied
cognitive sciences]” (p. 86).
But:
Ecological psychology explicitly rejects that “picking up” information involves gathering,
internalizing, and processing:

“[Perceptual information] could not be transmitted at all. But then I do not believe that the visual
system is a channel for transmitting signals from the retina to the brain. I believe it is a system for
sampling the ambient array. […] And that means that the observer’s brain cannot be compared
to a computer, or to a processor of information delivered to the brain” (J. J. Gibson,
1970/1982, p. 86)

“Ecological information cannot be transmitted: it is ambient and available, not something to put
over a channel; it is something to be detected or used (or not) in regulating action. […]
Information pick up is not a process of “internalizing” information” (Reed, 1996, p. 155)

Perception depends on tracking specific patterns in the sensory array and exploiting
them to coordinate action.
3. Enactivim’s notion of “sense-making” vindicates constructivism
and representationalism

Enactivists are “in favor of interpreting the activity of perceptual agents as a

kind of construction of perceptually meaningful world” (Fultot et al.,
2014)

Enactivism “is germane to the representationalist, not ecological, theory of

cognition” (p. 304)

The same criticism can be found in: Hutto & Myin (2017), De Jesus (2016),
Oyama (2011), Riegler (2015)
But:

Some interpretations of “sense-making” steer clear of the constructivist

interpretation:

“Sense-making is the capacity of an autonomous system to

adaptively regulate its operation and its relation to the environment
depending on the virtual consequences for its own viability as a form of
life. Being a sense-maker implies an ongoing (often imperfect and
variable) tuning to the world and a readiness for action.” (Di Paolo
et al., 2018, p. 33)
*Special thanks to: Ezequiel Di Paolo, Xabier Barandiaran, Nick Brancazio.

Thank you!
Miguel Segundo Ortin
www.miguelsegundoortin.com