Available online at www.sciencedirect.

com
Consciousness
and
Cognition
Consciousness and Cognition 17 (2008) 1324–1340
www.elsevier.com/locate/concog

Action-dependent perceptual invariants: From ecological

to sensorimotor approaches
Matteo Mossio a,b,1,*, Dario Taraborelli c
a
IHPST, CNRS/Université Paris I, 13 rue du Four, 75006 Paris, France
b
CenECC, École Normale Supérieure, 45 rue d’Ulm, 75005 Paris, France
c
Department of Psychology, University College London, Gower Street, London WC1E 6BT, UK

Received 26 March 2007

Available online 28 January 2008

Abstract

Ecological and sensorimotor theories of perception build on the notion of action-dependent invariants as the basic struc-
tures underlying perceptual capacities. In this paper we contrast the assumptions these theories make on the nature of per-
ceptual information modulated by action. By focusing on the question, how movement specifies perceptual information, we
show that ecological and sensorimotor theories endorse substantially different views about the role of action in perception.
In particular we argue that ecological invariants are characterized with reference to transformations produced in the
sensory array by movement: such invariants are transformation-specific but do not imply motor-specificity. In contrast,
sensorimotor theories assume that perceptual invariants are intrinsically tied to specific movements. We show that this
difference leads to different empirical predictions and we submit that the distinction between motor equivalence and
motor-specificity needs further clarification in order to provide a more constrained account of action/perception relations.
Ó 2007 Elsevier Inc. All rights reserved.

Keywords: Action; Perception; Self-motion; Invariants; Sensorimotor theories; Ecological psychology; Enaction

1. Blurring the boundary between action and perception

The idea that action and perception are not two independent cognitive domains and that perception is consti-
tutively shaped by action has been an important research trend in the last decades. The study of active perception,
or those mechanisms that allow organisms to negotiate perceptual problems by processing sensory stimulation
on the basis of action, has become a thriving area of investigation in neuroscientific, psychological and compu-
tational research on perception, witness the number of reviews and theoretical analyses that have addressed this
trend (Berthoz, 2000; Findlay & Gilchrist, 2003; Hurley, 1998; Port & Van Gelder, 1995; Thelen & Smith, 1994).
The idea of perception as action-dependent has been particularly emphasized by motor theories of percep-
tion, i.e. those approaches claiming that perceptual content depends in an essential way on the joint contribu-

*
Corresponding author. Fax: +33 1 44 27 86 47.
E-mail address: [email protected] (M. Mossio).
1
IHPST, CNRS/Université Paris I, 13 rue du Four, 75006 Paris, France.

1053-8100/$ - see front matter Ó 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.concog.2007.12.003
 M. Mossio, D. Taraborelli / Consciousness and Cognition 17 (2008) 1324–1340 1325

tion of sensory and motor determinations (Sheerer, 1984). According to motor theories, perceptual systems
are able to discriminate between reafference (sensory input resulting from self-motion) and exafference (sen-
sory input produced by external events) in virtue of their relation to efference (internal information elicited
by self-motion). Perceptual systems receive at the same time reafference and efferent copies generated by a
given movement and such information is used to perceive a specific action as self-initiated. Over time, the
organism learns to establish correlations (supposedly stable and systematic for a given organism in a given
environment) between reafference and efference (Gallistel, 1980). The capacity to discriminate between exaf-
ference and reafference underlies, according to motor theories of perception, several perceptual distinctions
(e.g. self-motion vs. motion in or of the environment) and constancies (e.g. position constancy, see Duhamel,
Colby, & Goldberg, 1992; Stark & Bridgeman, 1983).
Motor theories of perception represent a liberal view of action/perception relations, to the extent that they
admit that perceptual capacities can depend on sensorimotor relations on top of purely sensory information
(Hurley, 2001). They are mildly liberal, though, insofar as they limit the contribution of action to perception to
those perceptual processes that underlie the control of motor behavior. This is consistent with the established
view that part of the human perceptual system, geared towards action control and spatial behavior, draws on
dedicated processes that are functionally independent from those targeted at perceptual categorization and
recognition (Jacob & Jeannerod, 2003; Milner & Goodale, 1995). But can action-dependent perceptual pro-
cessing extend beyond what seems to be its natural domain, i.e. perception for motor control and spatial
behavior? Can action modulate a broader range of perceptual processes than mainstream perceptual theories
assume? Does the perception of properties of the environment (independent on the agent’s body configuration
or movement) rely on action-dependent information? This is a much stronger claim of action-dependence in
perception and an issue we aim to tackle in this paper.
A number of research programs have taken a more radical stance on action/perception relations, claiming
that action is pervasive in the functioning of perceptual systems. The role of action in perception, they suggest,
extends beyond processes targeted at the control of motor behavior and some aspects of perceptual perfor-
mance intrinsically depend on the contribution of action.
Claiming that perception is intrinsically active, inseparable from action or even—as some have argued—
reducible to a form of action (Noë, 2004), is a controversial hypothesis raising different orders of questions.
At a conceptual level, this hypothesis questions the traditional distinction between perception and action as
independent cognitive domains. A lively debate has addressed this claim, and several lines of criticism have
been leveled against the claim that perception cannot occur without action (Jacob, 2006; Prinz, 2006). Not
surprisingly, more attention has been paid to the theoretical implications of these radical approaches than
to the consequences of implementing a radical approach as an empirical research program.
From an empirical perspective, radical approaches on action/perception relations raise an interesting, often
understated question. If part of perception is action-dependent, what processes and structures can be invoked
to explain how perceptual systems parse and select sensory information on the basis of action? The answer
that these theories seem to suggest is that perceptual systems process action-dependent perceptual invariants,
or patterns of perceptual information that are intrinsically action-dependent.
In this paper our aim is not to address the question whether action and perception can be considered as
mutually independent or to argue in favor or against radical approaches. The goal of our analysis is to artic-
ulate the very idea that perception may rely on action-dependent invariants and to contrast the way in which
such a notion is characterized in these radical approaches. We propose that an analysis of this notion is a more
promising testbed to assess the prospects of radical approaches to action/perception as genuine empirical pro-
grams than a generic criticism of some of their (possibly flawed) background assumptions.

1.1. Radical approaches to action/perception relations

The most prominent theories that take a radical view of the contribution of action to perceptual processes are
those that belong to the ecological approach to perception (Cutting, 1986; Gibson, 1979; Reed, 1996). The ecolog-
ical approach emphasizes the constitutively active nature of perceptual abilities, and the fact that perceptually
relevant information is revealed by active interaction of the observer with the environment. A review of the eco-
1326 M. Mossio, D. Taraborelli / Consciousness and Cognition 17 (2008) 1324–1340

logical tradition is beyond the scope of the present analysis. Here we focus on ecological theories as a case of rad-
ical view on action/perception relations in order to analyze the notion of perceptual information they endorse.
The contrast class that we consider in this paper is a family of research programs that, although not orga-
nized into a unified theoretical framework, share a strong interactivist assumption about the nature of percep-
tion. We refer to this class of programs as the sensorimotor approach to perception. The sensorimotor approach
puts a strong emphasis on the fact that the basic structures underpinning perceptual skills are couplings
between an organism’s movements and co-occurring changes in sensory stimulation. In this sense, sensorimo-
tor approaches are a radicalization of classical motor theories of perception insofar as they emphasize the per-
vasiveness of information about self-initiated movement in perceptual processes beyond those targeted at the
control of motor behavior.
Although this assumption is not yet articulated into a full-fledged empirical paradigm (such as the ecolog-
ical approach to perception), a long tradition supports the idea that sensorimotor couplings are key to the
understanding of perceptual phenomena. The first explicit acknowledgement of the role of sensorimotor invar-
iants to understand perception can be traced back to Helmholtz, who suggested that:
[w]hen we perceive before us the objects distributed in space, this perception is the acknowledgement of a
lawlike connection between our movements and the therewith occurring sensations [...]. What we per-
ceive directly is only this law (Helmholtz, 1878/1977, pp. 138–139).
Sensorimotricity is a core notion in enactive theories, which take cognition as a capacity resulting from the
self-organization of living systems and from the constraints imposed on this activity by the interaction with the
external environment (Jarvilehto, 1999; Maturana & Varela, 1980; Varela, 1979; Varela, Thompson, & Rosch,
1991). In the field of neurophysiology, some authors defended the idea that perception reduces to the ability of
encoding patterns of covariation between motor patterns and co-occurring sensory stimulation (MacKay,
1962, 1987). Building on inspiration that can be traced back to Poincaré (1902), sensorimotricity as the ability
to process systematic couplings between sensory and motor information has also been regarded as constitutive
to the acquisition of space perception (Philipona, O’Regan, & Nadal, 2003; Wolff, 2004). A sensorimotor
hypothesis has been recently applied to the study of color categorization, showing that several psychophysical
aspects of color perception can be predicted by looking at how invariants in properties of reflecting surfaces
are encoded by the perceptual system of an agent actively exploring its environment (Philipona & O’Regan,
2006). Finally, and more controversially, sensorimotricity has also been proposed as a key to the explanation
of the phenomenal character of visual experience and visual consciousness (O’Regan & Noë, 2001).
Ecological and sensorimotor approaches have often been conflated or considered variations on the same
theory (Pylyshyn, 2001; Scholl & Simons, 2001) insofar as they share an important number of background
assumptions, in particular:

A. The claim that many traditional problems in perceptual theory not taking into account the contribution
of action (e.g. stimulus disambiguation, inverse optics problems, perceptual binding), are ill-posed.
B. The idea that perceptually relevant patterns of the sensory stimulation are those selected through motion
and that static sensory patterns (e.g. properties of the retinal image) are irrelevant to functionally char-
acterize perception.
C. The claim that perceiving does not require possessing detailed internal representations of the external
environment.
D. The focus on the intrinsically active nature of perception and the role of action-perception loops.

These similarities have often been taken as arguments in support of a substantial theoretical continuity
between ecological and sensorimotor approaches as opposed to mainstream perceptual theories.1 Defen-
dants of a sensorimotor approach have described their divergence from Gibsonian theories as a mere matter
of ‘‘explanatory focus” and denied any major theoretical discontinuity between the two approaches (Noë,
2002):

1
See the debate hosted at http://www.interdisciplines.org/enaction.
 M. Mossio, D. Taraborelli / Consciousness and Cognition 17 (2008) 1324–1340 1327

In adopting this view we invoke the role of action and the importance of extracting invariants, and so we
are indebted to Gibson. But we harness these ideas for quite different explanatory and theoretical purposes.
In other words, whereas Gibson stresses the use of sensorimotor invariants as sources of information, we
are stressing the idea that sensorimotor invariants are part of what constitute sensations and perceptual
content. We show that Gibson’s idea can go farther than Gibson pushed it (O’Regan & Noë, 2001, p. 1019).

Similarly, defendants of the ecological approach to perception tend to present their theories as an articu-
lation of the enactive approach to perception, fully compatible with a sensorimotor approach (Stoffregen &
Bardy, 2004; Stoffregen, Bardy, & Mantel, 2006). In spite of the large number of shared background assump-
tions, the continuity between these approaches has been challenged by other authors (Hurley, 2001; Varela
et al., 1991). What this debate has failed to appreciate is the difference in the specific constraints that these
theories put on perceptual information modulated by action. The thesis we defend in the present work is that
of a substantial divergence between ecological and sensorimotor approaches to perception: this divergence
may not be explicit at a broad theoretical level, but—we argue—underpins specific predictions made by these
theories on how action modulates perception.
Our argument is twofolded. On the one hand, we will contrast ecological and sensorimotor approaches to per-
ception by looking at the way in which they characterize the notion action-dependent information. On the other
hand, we will argue that, because of their different characterization of perceptual invariants, these approaches
make different empirical predictions about movement-mediated perceptual skills and about perceptual informa-
tion and processes underpinning such skills. We conclude that this distinction between different notions of invari-
ant is crucial to frame any empirical research program on the functional role action plays in perception.

2. Framing the concept of action-dependent invariants

The radical nature of ecological and sensorimotor approaches to perception does not simply consist in the
claim that perception must be studied by looking at the interaction between a goal-directed, active organism
and its environment.2 A much stronger claim they make is that information grounding perceptual abilities is
itself action-dependent. In this sense, referring to modulations of perception through action requires formu-
lating explicit hypotheses on how movement affects perceptual information. In this section, we review the
characterizations of the notion of action-dependent invariants that can be found in the ecological and senso-
rimotor literature and we show to what extent they diverge.

2.1. Ecological invariants

An informal characterization of the notion of a perceptual invariant can be found in J. Gibson’s seminal
work (Gibson, 1959, 1960). By providing a plethora of examples of perceptual structures that can be extracted
in action/perception couplings, Gibson offers a sort of extensional characterization of what an action-depen-
dent perceptual invariant may be. The lack of an explicit characterization of what constitutes action-depen-
dent perceptual information has been one of the most criticized aspects of Gibson’s theory of perception
(Fodor & Pylyshyn, 1981) and later defendants of the ecological approach to perception have tried to spell
out this notion in more precise terms.
A more accurate characterization can be found in Michaels and Carello (1981) who provide a number of
definitions of perceptual invariants. In particular, they distinguish between structural invariants and transfor-
mational invariants. On the one hand, structural invariants are defined as those properties of the sensory stim-
ulation that remain constant through motor interaction, even though other properties may vary: as such they
determine classes of equivalence that allow distinct perceptual objects to be regarded as the same under this
respect. On the other hand, transformational invariants can be characterized as ‘‘modes of change” of percep-
tual objects, i.e. invariant dynamics of sensory stimulation produced by specific transformations on these
objects. We might rephrase this by saying that structural invariants are those properties that allow perceptual
2
See Section 2.4 below.
1328 M. Mossio, D. Taraborelli / Consciousness and Cognition 17 (2008) 1324–1340

systems to parse structural components of the environment, whereas transformational invariants allow per-
ceptual systems to detect and track dynamic regularities to which structural components obey.
The distinction between structural and transformational invariants has been criticized by other exponents
of the ecological approach (see for instance Cutting, cit., p. 67).3 However, a common feature of all these char-
acterizations, that remains faithful to a Gibsonian view, is the idea that invariants are patterns in sensory
information that are revealed when an organism engages in motor interaction with the environment, or:
structures that remain invariant despite certain transformations caused by the animal and that therefore
might serve to specify persisting environmental resources (Reed, cit., p. 48).
Let us try to put this notion of invariant in the context of the general assumptions of ecological theories of
perception. The ecological approach defends the idea that there is a nomological relation between specific
states of the agent–environment system and invariant properties of the proximal stimulation they produce
on the sensory organs of the perceiver. It is in virtue of this nomological relation—described by the laws of
ecological optics, acoustics, haptics. . . (Gibson, 1979)—that invariants ‘‘specify” or carry reliable information
about the states of the agent–environment system they refer to. This relation of specification in virtue of eco-
logical laws is the tenet of ecological accounts of perception:
[i]n ecological optics we typically assume that the information (the invariant) specifies the object or event
perceived; that is, we pick up (process) the invariant, and as a result perceive the object properties spec-
ified (Cutting, cit., p. 71).
Ecological theories assume that perceptual systems are geared to these proximal invariants insofar as their
‘‘pickup” enables the perception of states of the agent–environment system.4 More precisely, the ecological
approach postulates that perceptual systems rely on information provided by so-called ‘‘ambient energetic
arrays” (Gibson, 1979). By hypothesis, these energetic arrays provide information that is rich enough to spec-
ify all relevant properties of the agent–environment system and to perceptually control behavior.5 In partic-
ular, the ambient array contains multimodal sensory invariants that the agent can rely upon to perceive its
body configuration and movements (proprioception), the relation between itself and the environment, for
instance in order to determine its position or motion direction (exproprioception), as well as to perceive prop-
erties of the external environment (exteroception).
All of these proximal invariants are made available through motor interaction. Since a transformation is
needed to reveal the invariant, the availability of perceptual information necessarily requires action, as a condi-
tion to submit the sensory stimulation to the appropriate transformations. In this hypothesis lies one of the rad-
ical claims ecological theories make with respect to motor action: action is a necessary requirement to obtain
perceptually relevant information, and no perceptual ability can occur if invariants specified by action are not
available. To clarify the notion of perceptual invariant in the ecological approach, let us review two classical
examples.
A first classical example of perceptual invariant revealed through motor transformations is the so-called
cross ratio, a visual invariant specifying rigidity (see Fig. 1).6

3
Cutting (cit.) proposes the following characterization of an optical invariant: ‘‘To be an optic invariant, all information about an object or
event must be present in the optic array, measurable at a particular place and time, and valid to all places and times. Thus the invariant is a
constant mapping from the proximal image and the distant stimulus, where relations between image (or eye) and stimulus are not fixed” (p. 75).
4
See for instance Lee (1976) in the case of vision and Turvey (1996) for the dynamic touch.
5
As Warren (2006b) characterize it, ‘‘[i]nformation consists of patterns of stimulation at the receptors that are specific to the ecological
state of affairs and are therefore useful in controlling action” (p. 367).
6
As Cutting explains it: let A, B, C and D be four points on the same straight line (L1). Let X be a point not on that line, and connect all
point to X. This creates the new lines of which AX, BX, CX and DX are segments. Let line L2 intersect these new lines at points A0 , B0 , C0
and D0 . Projective geometry tells us that the cross ratios of segments bounded by the points ABCD and A0 B0 C0 D0 are the same. In
particular, the following segments lengths form the following equal ratios:
ðAD  BCÞ=ðAC  BDÞ ¼ ðA0 D0  B0 C 0 Þ=ðA0 C 0  B0 D0 Þ
This cross ratio—the product of the longest segment AD and the inner segment BC divided by the product of the segments connecting non-
adjacent exterior and interior pairs of points (AC and BD)—is invariant under any projection to any point not aligned with A through D
(Cutting, cit., p. 81).
 M. Mossio, D. Taraborelli / Consciousness and Cognition 17 (2008) 1324–1340 1329

Fig. 1. The geometric definition of cross ratio.

Two fundamental properties of the cross ratio need to be emphasized. First, cross ratios are invariant under
all rotations and translations of line L2, translations of line L1 and of point X. Second, it can be proved that
cross ratio invariance is preserved whatever the shape of the projection surface (L1) is. The invariance of cross
ratio under several transformations affecting its geometrical projection makes it a good candidate—from an
ecological perspective—as a perceptual invariant. Cross ratios can be taken as information that perceptually
specifies object rigidity in our environment.7 In this sense, cross ratio is a paradigmatic example of a geometric
invariant that remains unchanged throughout transformation in the proximal sensory stimulation of an obser-
ver and that bears a reliable informative relation with a property of the distal environment.
A second paradigmatic example of invariant addressed in the ecological literature is motion parallax, i.e. the
optical pattern produced by the relative movement of an observer with respect to objects in the visual envi-
ronment (Gibson, 1950). When an observer moves in space, the resulting displacement of the point of fixation
generates different motion of pairs of points in its visual stimulation due to the different distance of distal
objects from the fixation point (see Fig. 2).
Motion parallax can be analyzed in terms of several differential invariants: divergence, curl and deformation
(Koenderink & Van Doorn, 1975; Koenderink, 1986).8 It has been shown that these invariants provide reliable
perceptual information about the spatial structure of the stimulus, and provide in particular reliable informa-
tion for depth perception. Accordingly, a large number of empirical studies have explored humans’ capacity to
perceive depth (including objects three-dimensional shape) by relying uniquely on motion parallax, a process
called structure-from-motion (Nawrot, 2003; Ono, Rivest, & Ono, 1986; Rogers & Graham, 1979; Steinbach,
Ono, & Wolf, 1991; Ujike & Ono, 2001). It has been shown in particular that humans can use motion parallax
to perceive depth in passive conditions involving object motion (the ‘‘kinetic effect”, see Wallach & O’Connell,

7
It should be noted that merely extracting invariant cross ratios cannot explain the perception of rigidity in complex cases, for which
further perceptual information is required (Koenderink, 1986).
8
These invariants are referred to as ‘‘first-order differential invariants” of optic flow because their values are independent of both the
choice of the coordinate system and any rotations of the observer around the projection center. As Koenderink (1986) clarifies: ‘‘The
[divergence] is a number that specifies the relative time change of apparent area (solid angle) of a piece of the optic array, the curl is a
number that specifies the rate of rotation, and the [deformation] can be specified with a number (the degree of shear: always positive) and
an orientation (the axis of contraction)”.
1330 M. Mossio, D. Taraborelli / Consciousness and Cognition 17 (2008) 1324–1340

Fig. 2. Motion parallax. As the eye moves from left to right, closer points move faster than further points.

1953), as well in conditions involving either object motion or self-motion (Rogers & Graham, 1979, 1982; Wal-
lach, Stanton, & Becker, 1974).
It should be noted that the concept of a perceptual invariant as characterized in the ecological approach is
not limited to specific sensory modalities, as vision. Invariants can be both modal and multimodal and depend
on proximal information made available to the agent via the somatosensory and vestibular systems on top of
other sensory modalities. For instance, an important number of studies have tackled the issue of ‘‘dynamic
touch”, the ability of the agent to extract invariant properties of rotational dynamics (the inertia tensor),
which may specify objects and body properties (Carello, Kinsella-Shaw, Amazeen, & Turvey, 2006; Carello
& Turvey, 2000; Turvey, 1996).
The above examples illustrate the idea according to which perceptual capacities rely on invariants that the
agent can extract from a multimodal array of sensory information when this sensory information is modulated
by motor behavior. To summarize, we can say that in the ecological approach, a perceptual invariant is a
property of the proximal sensory array that remains unchanged throughout motor interaction with the envi-
ronment. It is action-dependent insofar as it can only be revealed when the sensory stimulation undergoes
transformation, which is typically the result of motor behavior.

2.2. Sensorimotor invariants

The idea that the joint effects of self-initiated movement and sensory stimulation can provide information
to perceptual systems is the core assumption of sensorimotor approaches to perception. Sensorimotor invar-
iants—often referred to as ‘‘sensorimotor contingencies” (MacKay, 1987; O’Regan & Noë, cit.) or ‘‘sensori-
motor dependencies” (Broackes, 2001; Philipona et al., 2003)—are the basic constituents of action-dependent
perceptual information in sensorimotor theories. O’Regan & Noë (cit.), propose the following informal char-
acterization of sensorimotor contingencies:

[t]he structure of the rules governing the sensory changes produced by various motor actions, that is,
what we call the sensorimotor contingencies governing visual exploration (p. 941).

This definition is quite loose, and in any case not sufficiently constrained to let us appreciate any significant
difference between invariants studied from a sensorimotor or ecological perspective. A better characterization
of a sensorimotor invariant can be found in the formal definition of ‘‘sensorimotor laws” proposed by
 M. Mossio, D. Taraborelli / Consciousness and Cognition 17 (2008) 1324–1340 1331

Philipona, O’Regan, Nadal, and Coenen (2004). Their goal is to characterize the invariance governing rela-
tions between motor efference (M) and sensory reafference (S). Given the configuration of the body (referred
to as P) and that of the environment (E), they propose that:

 P and M are connected through a function ua: P = ua(M)

 S is connected to E in virtue of a function ub: S = ub(P, E)

The following relation is then introduced as a functional sensorimotor law:

/ðM; EÞ ¼ /b ð/a ðMÞ; EÞ

Although this formal characterization can be challenged on some of its underlying assumptions,9 it is suffi-
ciently explicit to contrast the notion of a sensorimotor invariant from that of an ecological invariant. Senso-
rimotor laws constrain the way in which motor efference and sensory reafference systematically co-vary, as a
function of both the configuration of the body and the structure of the environment. In a given environment,
the body configuration is controlled by the motor outputs in virtue of ua. The sensory organs altogether
deliver a multidimensional input that is, in turn, a function ub of the configuration of the body and the
configuration of the environment.
Given the existence of these nomological relations in the agent–environment system, sensorimotor
approaches assume that perceptual systems are able to extract invariant relations between co-occurring sen-
sory and motor patterns. Sensorimotor invariants are by hypothesis the basic constituents of any kind of
information perceptual systems are able to parse when an organism interacts with the environment. In this
sense, sensorimotor theories endorse the controversial view that all perceptual information is inherently effer-
ence-dependent.
Given these considerations, we can now formulate a sufficiently distinctive characterization of an action-
dependent perceptual invariant from a sensorimotor perspective. In contrast to ecological approaches, senso-
rimotor theories do not characterize perceptual invariants as properties of the sensory stimulation revealed
through motion. A perceptual invariant in a sensorimotor approach is a property of sensorimotor couplings
that remains invariant throughout transformations. Whenever a sensory transformation co-occurs with a
movement, those properties that remain unchanged in this coupling constitute a sensorimotor invariant. Sen-
sorimotor invariants are then action-dependent insofar as they describe sensory transformations matching
specific movements and, vice versa, motor transformations matching specific sensory changes.

2.3. Ecological vs. sensorimotor invariants

As noted in the beginning of our analysis, the radical character of sensorimotor and ecological approaches
does not consist in the claim that action is required in the perception of one’s own movement, bodily config-
uration or, more broadly, in perception targeted at the control of spatial behavior (which is plainly compatible
with mainstream perceptual theories). Rather, the radical claim consists in saying that even the perception of
(supposedly) agent-independent properties (typically: exteroception) relies on action-dependent information.
Our thesis is that radical approaches diverge in their interpretation of the claim that action pervasively mod-
ulates perception, since they provide substantially different characterizations of what constitutes perceptual
information. In particular, we have shown that ecological and sensorimotor approaches endorse different
notions of perceptual invariants, and, more specifically, that they make contrasting assumptions on the rela-
tion of a perceptual invariant to the agent’s movement.
We submit that, in an ecological framework, it is not necessary to refer to specific movements in order to
characterize perceptual invariants but only to properties of sensory patterns that remain invariant across
possible transformations. In contrast, in a sensorimotor framework, invariants are by definition properties
of co-occurring sensory reafference and motor efference. More generally, we can say that movement has a

9
For instance, body configurations need not be uniquely a function of motor states but may also depend on environmental states (e.g.
orientation with respect to gravity).
1332 M. Mossio, D. Taraborelli / Consciousness and Cognition 17 (2008) 1324–1340

constitutive role in the determination of sensorimotor invariants, whereas it is instrumental in the determina-
tion of ecological invariants.
Let us try to unpack this idea. From an ecological perspective, properties of the sensory stimulation that do
not change under transformation are taken as perceptual invariants and as bearers of information about states
of the agent–environment system, in virtue of the ecological laws. Hence, picking up such invariants enables
the observer to perceive a given state, event or property related to the agent, the environment or their mutual
relation. If an ecological invariant is described by specific transformations in the multisensory array, move-
ments of the agent are only required to produce the sensory transformation with respect to which the invariant
is defined. However, in terms of ecological laws, multiple movements can in principle produce identical trans-
formations and reveal the same invariants. We then conclude that ecological invariants are defined in the gen-
eral case with respect to a specific transformation, but not to the movements that produce these
transformations.
It should be emphasized that the instrumental role of movement in constituting ecological invariants is not
at odds with the fact that they can provide information about the agent’s body configuration and movements
when needed. Perceptual invariants encompassing, for instance, somatosensory and kinesthetic stimulation
can specify proprioceptive exproprioceptive and/or exteroceptive information (the literature about dynamic
touch we mentioned before is particularly relevant here). However, from the fact that these sensory sources
can provide agent-related information, it does not follow that the underlying invariants need to encode move-
ments (and, in particular, efference) as constitutive variables.10
Consistently with the idea of the instrumental role of movement, the ecological approach admits the pos-
sibility that, in many perceptual conditions where information about the agent’s movements or body config-
uration is not necessary (typically, in cases of exteroceptive perception), invariants not encompassing agent-
related variables could provide sufficient information to accomplish the perceptual task. As we will clarify
in the following sections, distinguishing between the constitutive role of movement in invariants and the fact
that sensory invariants can inform the agent about its own movements has crucial implications on the explan-
atory strategy of the two approaches.
The idea of the instrumental role of movement in ecological approaches can be exemplified by considering
the two cases of perceptual invariants described above.
The cross ratio is invariant under different transformations affecting either the viewpoint X or the object
in the environment (L1). This means that, in real-world situations, the same invariant would be revealed
either by active movements of the perceiver around a rigid stationary object or by a moving object for
a static observer. Even if we consider active movements of the perceiver, the same invariant can be revealed
either by the active manipulation of the object by the perceiver or by an active displacement of his view-
point. Since what matters from an ecological perspective is the informative relation between the perceptual
invariant (e.g. the cross ratio) and the specified property of the distal object (e.g. rigidity), the specific
motor patterns producing the sensory transformation do not contribute to the characterization of the
invariant.
Similarly, what specifies motion parallax as an invariant is a transformation that underdetermines the class
of possible movements that produce it. In order for parallax to be revealed, the relevant sensory transforma-
tion has to occur, but it makes no difference, in terms of the availability of the information, if the invariant is
revealed by an active movement of the observer, by a passive movement of the observer, or by a movement of
the object in the environment.
In contrast, according to a sensorimotor approach, when actively exploring the environment an organism is
exposed to dynamic sensorimotor couplings, which possess properties invariant under transformations affect-
ing various efferent motor and reafferent sensory variables. For instance, invariant relations can be described
between a rotation of the head on the one hand and the co-occurring patterns of retinal and proprioceptive
stimulation on the other hand. These sensorimotor relations remain invariant under transformations affecting
sensory or motor parameters (i.e. changes in rotation velocity, changes in initial conditions, variations in

10
Incidentally, different sources of motor-dependent information can fail to uniquely specify the state of the agent, see for example the
discussion about input conflicts in Stoffregen and Bardy (2001).
 M. Mossio, D. Taraborelli / Consciousness and Cognition 17 (2008) 1324–1340 1333

stimulus configuration, or intensity). The perceiver is able by hypothesis to identify sensorimotor invariants
over several occurrences of sensorimotor couplings.
We can summarize these conclusions by saying that these two characterizations of the notion of action-
dependent perceptual invariants put different constraints on the kind of information an organism is suppos-
edly able to extract and to process. In the case of sensorimotor approaches, the organism is supposed to be
able of extracting motor-specific information whereas in the case of ecological theories the organism must
be able to extract transformation-specific information.

2.4. Motor-dependence vs. action-dependence

Are we entitled to conclude that ecological invariants, given their instrumental relation with movement, are
not action-dependent? This claim may sound counterintuitive, since the fundamental assumption of the eco-
logical approach is admittedly that perceptual information is only available to agents actively exploring their
environment. Perceptual information is available to agents in terms of affordances, i.e. it specifies ‘‘possibilities
for action” (Gibson, cit.; Withagen & Michaels, 2005). Invariants specify information only to animals engag-
ing in actions and in this sense information provided by invariants has been considered as constitutively action-
dependent (Gibson, 1979; Reed, 1996; Warren, 2006b). As Michaels and Carello (cit.) observe:
[e]nergy patterns that are invariant with respect to relevant transformations and that specify the environ-
ment are not, by themselves, equivalent to information; the animal must be specified. Moreover, even a
complete inventory of those invariant structures must be based on an equally complete inventory of
the consequences of each of the exploratory (and performatory) activities in which an animal can engage
(p. 39).
How can we accommodate our claim of the instrumental role of motion in ecological approaches with the
claim that action is constitutive of perceptual information according to defendants of ecological theories?
The solution, we argue, is that in the ecological approach the term ‘‘action” refers to adaptive, goal-directed
behavior (Gibson, cit.; Reed, cit.). By picking up perceptual invariants, organisms perceive affordances that
mediate adaptive control of behavior. In ecological terms then, the radical claim that information is ‘‘con-
stitutively action-dependent” means that animals perceive affordances in the environment that allow them to
attain specific goals.11 Information in terms of affordances is in this sense targeted to specific actions, but the
invariants (whose pickup reveals an affordance) do not depend, as we argued, on specific movements. As
Reed puts it:
[i]nformation is specific to the ecological task of the animal. It is not specific to mechanisms or to pro-
cesses within the animal, nor is it purely external unrelated to the organism (Reed, 1996, p. 57).12
Ecological approaches can then at the same time defend a broad idea of action-dependence of information
(i.e. its specificity to the task) without discarding the idea of the instrumentality of movement at the level of
perceptual invariants. As Withagen and Michaels (2005) point out, Gibson’s (1966) concept of a ‘‘perceptual
system” as well as Reed’s (1982) concept of an ‘‘action system” assume the functional dependence of action
and perception at the level of goal-directed behavior. The same information can be ‘‘captured” by different
perceptual systems which may physically vary, and cannot be described with reference to specific anatomical
structures (Reed, 1986).

11
As Warren (2006b) puts it: ‘‘Alongside the dynamical approach to movement there developed the ecological perception–action
approach to the control of behavior (Gibson, 1958/1998, 1979; Lee, 1976, 1980; Shaw, Kugler, & Kinsella-Shaw, 1990; Turvey & Carello,
1986; Warren, 1988, 1998). This view emphasizes the role of occurrent information in guiding behavior, in the form of optic, acoustic,
haptic, or olfactory fields that are structured by and are specific to the state of the agent–environment system. The research program
involves determining what informational quantities govern naturalistic behaviors like reaching, catching, hitting, standing posture, or
locomotion”.
12
By taking the classical example of s (Lee, 1980), Reed explains that ‘‘The hummingbird’s ability to maintain s > 0.5 is an ability to dock
with a slight closing velocity. Bats also seem to control locomotion by monitoring s. Despite the radical difference in mechanisms (vision
versus echolocation, bird versus bat flight patterns), these animals are accomplishing the same thing” (Reed, 1996, p. 57).
1334 M. Mossio, D. Taraborelli / Consciousness and Cognition 17 (2008) 1324–1340

The instrumental role of movement in the definition of invariants appears to be perfectly consistent with
this theoretical stance: ecological psychologists refer to this idea as ‘‘motor equivalence”. The ecological
approach insists on the idea that perceptual systems are able to exploit information which is invariant over
variations of body configuration, movements and even anatomical structure. This capacity has an adaptive
value, to the extent that it enables agents to obtain equivalent information under different conditions. Again,
emphasizing the idea of motor equivalence does not conflict with the fact that ambient arrays may provide
invariants specifying proprioceptive and exteroceptive information. Rather, it implies that agent-related infor-
mation will only be used when relevant with respect to the perceptual task, but neglected when irrelevant, con-
sistently with the idea of the adaptive value of motor equivalence.
To conclude, if we take action to mean ‘‘goal-directed motor behavior”, then we can say that the sensori-
motor approach shares with the ecological approach the assumption that information is action-dependent in a
loose sense.13 However, we argued that sensorimotor and ecological theories make contrasting assumptions on
the nature of perceptual invariants that constitute information for the agent. Albeit loosely ‘‘action-depen-
dent” in both cases, we suggested that perceptual information is characterized as motor-specific in the case
of sensorimotor invariants and transformation-specific (and motor equivalent with respect to movement) in
the case of ecological invariants.
This distinction between action-dependence and motor-dependence has been underestimated so far in the
literature. Part of the major theoretical divergence between enactive and ecological theories (in particular their
different view of agents as shaping the perceptual environment through action vs. agents simply exploring the
perceptual environment through action) should actually be reconsidered in the light of this technical distinc-
tion about the underlying notion of perceptual information.14
We submit, though, that this distinction is useful not only to distinguish the ecological and sensorimotor
approach at a theoretical level, but also to disentangle different empirical predictions that depend on this dis-
tinction and that the literature has not yet fully acknowledged.

3. Motor-specificity: empirical predictions of the sensorimotor hypothesis

If—as sensorimotor theories assume—perception is motor-dependent insofar as it systematically relies on

information that ties sensory reafference to specific classes of motor efference (and not only to motor-induced
sensory transformations), then we should expect a substantial divergence in the empirical predictions made by
ecological and sensorimotor theories.
The question of the empirical validation of these approaches reveals a major asymmetry between the sen-
sorimotor and the ecological theory. On the one hand, the ecological approach to perception can be seen as a
full-fledged psychological theory, which has received in four decades of investigations substantial empirical
support.15 Ecological psychologists have developed detailed descriptions of how perceptual systems exploit
different sources of information available in the ambient arrays for the adaptive control of behavior. On
the other hand, as we already pointed out, the sensorimotor approach has not yet been translated into an
empirical research program, and has been criticized precisely for the lack of empirical evidence in support
of its core assumption (Scholl & Simons, 2001). In particular, the thesis of the constitutive and pervasive role
of movement in perception appears to be an extremely strong claim, which, as many have suggested, might be
trivially refuted by cases in which accurate perceptual performance can occur with no motor contribution of
the agent (Prinz, 2006). The ecological perspective, by putting forward the idea that perceptual information is
motor-equivalent, seems to be offering a more plausible account of how to characterize the relations between
action and perception.
Moreover, one may argue that, even if the theoretical distinction between the constitutive and instrumental
role of movement in defining perceptual invariants holds, this makes virtually no difference between the two

13
‘‘Enactive knowledge depends upon an action-perception cycle. Action reveals information, which guides further action, which reveals
additional information, and so on” (Stoffregen & Bardy, 2004).
14
See also Varela et al., 1991, pp. 203–204 for a similar conclusion.
15
A representative bibliography of the theoretical and experimental work produced by the ecological approach can be found here: http://
ione.psy.uconn.edu/publications.html.
 M. Mossio, D. Taraborelli / Consciousness and Cognition 17 (2008) 1324–1340 1335

approaches, to the extent that ‘‘constitutively” motor-equivalent invariants may specify, when needed, agent-
related (e.g. proprioceptive and exproprioceptive) information. Hence, the ecological approach can account
for all perceptual tasks that require a reference to the agent’s self-motion, without postulating the pervasive-
ness of efference-dependent information.
In this section, we address the issue of the empirical basis by reviewing evidence from the experimental lit-
erature in which encoding motion specificity seems to be a key to the explanation of a number of perceptual
capabilities. The logic of this section consists in showing that there seems to be cases in which motor-specificity
plays a role in modulating perceptual judgments even if it does not provide additional information which is
relevant for the task. In other words, there are conditions in which efference-specific information seems to
be required not only to perform proprioceptive and exproprioceptive tasks targeted at the control of motor
behavior (in which case the ecological and sensorimotor approach would make similar predictions), but also
in the case of exteroceptive judgments where movement should not in principle contribute to further specify
the property or state to be perceived. These examples are not meant to be a rebuttal of the ecological thesis of
the functional equivalence of motor mechanisms, which has already received substantial empirical support (see
Section 2.1 above). Rather, our aim is to suggest that the sensorimotor notion of motor-specificity should be
carefully distinguished from that of transformation-specificity since the former could be in a better position
than the latter to account for some specific classes of perceptual abilities.

3.1. Active vs. passive perception

The first set of experimental results we review are those contrasting the performance of immobile vs. mov-
ing observers when they engage in perceptual tasks involving identical sensory configurations.

3.1.1. Perceptual substitution

Perceptual substitution is a manipulation that allows to transform stimuli in a given sensory modality (e.g.
vision) into stimuli of another sensory modality (e.g. touch). A perceptual substitution device consists of three
parts: a sensor allowing the conversion of a form of energy (stimuli from sensory modality A) into signals
interpretable by a coupling system, responsible of the coordinated activation of a stimulator into sensory
modality B. The goal of protocols using perceptual substitution devices consists in measuring to what extent
perceptual systems are able to correctly process information independently of the modal coding in which the
senses present it.
Experimental results obtained by perceptual substitution protocols (Bach y Rita, 1972; Barfield & Furness,
1995; Gonzalez & Bach-y-Rita, 2003; Lenay, Cannu, & Villon, 1997) support the claim that perceptual pro-
cessing required to solve these tasks rely on the availability of efferent information. Particular emphasis has
been put on the fact that efficient perceptual substitution only occurs if subjects are allowed to engage in sen-
sorimotor interaction with the object, whereas no perception arises if dynamical sensory stimulation is pas-
sively received and processed by the subject (Lenay, Gapenne, Hanneton, Marque, & Genouëlle, 2003;
Sampaio, 1995; White, Saunders, Scadden, Bach y Rita, & Collins, 1970). The interesting point is that, when
subjects are allowed to actively interact with the object, in many cases motor patterns do not reveal subject-
related sensory stimulation, which would provide additional relevant information for accomplishing the task.
To consistently account for these cases, the ecological approach should show that 1) only the active conditions
provide adequate information to the subject and 2) in the active conditions there is motor equivalence.16
Otherwise, from an ecological perspective one could expect a similar performance in both cases, since the rel-
evant invariants produced during the transformation of the stimuli are supposedly the same in the active and
passive conditions. In contrast, the sensorimotor approach straightforwardly predicts a different performance
in the two conditions, assuming that the observer, in the passive condition is not engaging in any kind of sen-
sorimotor coupling, and as a consequence no perceptual invariants can be extracted in this condition.

16
For instance, Warren (2006a) reports evidence that depth perception in sensory substitution is not motor specific, since it would
transfer from one arm to the other as well as to novel joint angles. He suggests that this is evidence of a perception-action invariant
relevant for sensory substitution that is not motor-specific.
1336 M. Mossio, D. Taraborelli / Consciousness and Cognition 17 (2008) 1324–1340

3.1.2. Structure-from-motion
In contrast with the results discussed in Section 2.1, which support the hypothesis of motor equivalence,
several studies have shown that subjects process identical visual configurations in different ways, depending
whether they actively produce the transformation in the visual stimulation or they passively process it (see
Wexler & Van Boxtel, 2005 for a comprehensive review). There is growing evidence that the perception of
depth and object shape from motion parallax considerably varies depending on whether the observer is
allowed to move or not (Van Boxtel, Wexler, & Droulez, 2003; Wexler, Lamouret, & Droulez, 2001b; Wexler,
Panerai, Lamouret, & Droulez, 2001a). Typical test conditions are those in which subjects are asked to dis-
ambiguate a 3D optic flow simulated on a computer monitor. The optic flow is presented in such a way that
the same ambiguous configurations are present when subjects passively experience the stimuli or when they are
allowed to move towards the object. Wexler et al. (2001a) show that subjects provide different interpretations
of the stimuli in the passive and active conditions, despite the fact that they are exposed to the very same visual
sensory transformations. Several hypotheses have been proposed to account for the reason why the perception
of the very same visual configuration varies when the subject is moving (Wolpert & Flanagan, 2001). All of
these hypotheses require that the agent be capable of predicting the correlation between specific efferent pat-
terns and specific sensory transformations. The difference between the two conditions cannot be explained
solely by referring to sensory transformations (Findlay & Gilchrist, 2003; Gibson, 1950). Again, the fact of
perceiving differently an identical optic flow seems to be at odds with the ecological assumption of motor
equivalence, since in this case self-motion does not reveal additional information supposed to be relevant
for the perceptual task. Conversely, if the difference in performance is due to different ways of processing sen-
sorimotor couplings, then sensorimotor invariants seem to be the relevant structures that agents are
extracting.
A defendant of the ecological approach may also reply that all these experimental protocols comparing
active vs. passive conditions do not provide any evidence specifically in favor of the sensorimotor hypothesis,
since the ecological approach would equally predict that, as far as in one condition the perceiver is not actively
exploring the environment, his performance will trivially be different from the active condition. Passive con-
ditions, as it were, do not provide information in terms of affordances because there is no goal-directed behav-
ior. However, insofar as the observer can make distinct perceptual judgments, we argue that it is legitimate to
consider information provided by the simulated optical flow as ‘‘affordance” in a genuine ecological meaning.
The fact the perceiver is not actively moving should make no difference if motor patterns coupled with the
sensory stimulation were irrelevant to the task.

3.2. Perception in active vs. passive self-motion

A second set of experimental results consistent with the hypothesis of motor-specificity of perceptual invar-
iants are those suggesting that subjects perceive differently identical dynamic visual configurations whether
their movement is self-generated (‘‘voluntary”) or passively imposed. For instance, it is know for a long time
that perceptual adaptation to prismatic distortions is much more efficient in conditions where movement is
self-generated than in conditions where participants are passively moved on a trolley or in a wheelchair (Held,
1965). Analogously, active self-motion has been shown to improve heading judgments when compared to pas-
sive self-motion (Telford, Howard, & Ohmi, 1995).
Wexler (2003) compares a control condition in which subjects experience an ambiguous dynamical config-
uration (which can be interpreted as stationary either in an allocentric frame of reference, or in an egocentric
frame of reference) when engaging in self-generated head movements with a condition in which subjects have
their head blocked and are passively displaced on a wheelchair by the experimenter. With respect to the active/
passive comparison, in this case subjects are able to move their head in both conditions, which provides them
with additional proprioceptive information. The results show a significant difference in the interpretation of
the visual configuration in the two conditions: subjects display a systematic preference for the allocentric sta-
tionary interpretation when head movements are self-generated. Crucially, Wexler shows in the same study
that perceptual performance changes not only as a function of the active or passive character of self-motion,
but also of the specific kind of active movement performed by the subject. Subjects presented with an identical
visual pattern are asked to perform the same task of stimulus disambiguation either by freely moving their
 M. Mossio, D. Taraborelli / Consciousness and Cognition 17 (2008) 1324–1340 1337

head or by actively moving the wheelchair without the possibility of moving their head. In this case, results
show that the interpretation of the visual stimuli was significantly different. Whereas subjects preferred again
an allocentric interpretation when allowed to actively move their head, they systematically preferred an ego-
centric interpretation when forced to move the wheelchair, thus showing the determinant role of the specific
motion scheme in modulating perceptual judgments.
The standard account for these results is that subjects perform differently in the two conditions because
active self-motion provides additional extra-retinal information with respect to passive self-motion. In partic-
ular, efferent copies are assumed to enable sensorimotor predictions—i.e. anticipations of sensory effects
resulting from specific movements—which may be used to constrain perceptual judgments through the com-
parison between expected reafference and actual sensations (Blakemore, Frith, & Wolpert, 1999; Wolpert &
Flanagan, 2001). In the specific case of ambiguous configurations discussed, sensorimotor prediction may
be used to constrain the possible interpretations of the visual configuration (Wexler & Van Boxtel, 2005).
As Wexler emphasizes, the hypothesis of sensorimotor predictions is consistent with the fact that the percep-
tual performance varies following the specific motor scheme in which the subject is engaged.
Evidence on the contribution of voluntary self-motion to perception represents a second example of empir-
ical results supporting the idea that ecological invariants that postulate motor equivalence cannot per se
account for cases in which different motion patterns coupled with the same sensory stimulation result in dif-
ference in performance. Sensorimotor theories would argue that the difference in the two conditions is due to
the use of distinct types of underlying motor-specific invariants, whereas ecological theories would need to jus-
tify the appeal to proprioceptive information as a necessary integration to disambiguate identical transforma-
tional invariants.

3.3. Can motor-specific perceptual invariants shed light on action/perception relations?

The results shortly reviewed in this section suggest that in several conditions only by assuming that subjects
are able to parse the specific movements in which they engage when involved in a perceptual task, it is possible
to adequately account for major differences in perceptual performance. Whether motor-specificity can be
extracted simply in virtue of underlying invariants or requires the integration of a proprioceptive judgment
based on non motor-specific information is an issue that requires further empirical investigation. These cases
are meant to illustrate that theories endorsing different notion of action-dependent perceptual invariants can
be contrasted not only at a conceptual level but also in terms of the different empirical predictions they for-
mulate on specific perceptual tasks. Of course, these results do not provide compelling evidence that percep-
tual systems, when negotiating stimulus interpretation through action, always rely on motor-specific
strategies. Quite on the contrary, the ecological hypothesis of functional equivalence of motor patterns is well
established and has received broad empirical support, which is in fact at odds with the predictions of senso-
rimotor theories. What this empirical evidence suggests is that the question of understanding to what extent
action modulates perceptual processing cannot be settled without a thorough understanding of the structure of
perceptual information mediated by action. For instance, an advocate of the ecological approach may argue
that these results simply suggest that, as far as the agent behaves in more ‘‘ecologically valid” conditions,
agent-related information provided by self-motion can contribute to exteroception in a broad range of condi-
tions. However, it seems that this general interpretation would significantly weaken (if not undermine) the
importance of the functional equivalence hypothesis, which is a distinctive and well-established feature of per-
ceptual systems according to the ecological approach.

4. Conclusions

The aim of the present analysis was to frame the notion of action-dependent perceptual invariants—a
notion that is central to controversial claims of the pervasiveness of action in the functioning of human per-
ception. By taking a radical stance on the functional role of action in perception, ecological and sensorimotor
theories are often considered as similar and mutually reducible theories. We argued that, in spite of a number
of shared background assumptions, ecological and sensorimotor approaches to perception make substantially
different predictions on the nature of perceptual information.
1338 M. Mossio, D. Taraborelli / Consciousness and Cognition 17 (2008) 1324–1340

We showed in particular that ecological theories conceive invariants as properties of the sensory stim-
ulation that remain constant through transformation produced by the perceiver’s self-motion, but in
which the actual movement generating the transformation is extrinsic to the characterization of the
invariant. Ecological invariants are transformation-specific, but they do not rely on the extraction of spe-
cific motor schemes: whatever motor scheme produces a transformation in the ambient array, only the
transformation is relevant to specify the perceptual invariant. In contrast, sensorimotor approaches pos-
tulate that perceptual systems build invariants by extracting regularities from co-occurring efferent motor
and reafferent sensory patterns. In this case, the perceptual system is supposed to be able to represent
invariances that are motor-specific, insofar as properties of motor schemes are intrinsic constituents of
perceptual invariants.
This distinction allows us to drive a wedge between theories that attribute an instrumental role to self-
motion in the specification of invariants and theories taking self-motion as constitutive to the specification
of an invariant. Far from being a purely conceptual distinction, we suggested that this difference results in
diverging empirical predictions about perceptual capabilities mediated by action. Whether an observer is
extracting regularities of her own motor schemes, has substantial consequences on how perceptual judgments
are affected by motor information.
Advocates of the ecological and sensorimotor approaches may still object that there is no conceptual
incompatibility between the hypotheses of motor equivalence and motor-specificity. Hence, each approach
could easily accommodate both hypotheses in its own framework. The ecological approach could (a) include
efferent information in the characterization of the global energetic array or (b) appeal to the contribution of
proprioception to account for cases in which simple transformational invariants are insufficient to ground per-
ceptual distinctions. Conversely, the sensorimotor approach may admit that (a) certain kinds of perceptual
abilities need just rely on transformation-dependent information or it may even accept that (b) the organism
must abstract from motor-specificity for certain classes of perceptual judgments to be possible. In this case the
ecological and the sensorimotor approaches would effectively reduce to the same theory. Nevertheless, such
unification does not come for free. If action-dependence in perception can alternatively be motor-equivalent
or motor-specific, the theory will need to provide predictions on those conditions in which action modulates
perception in a motor-specific way (e.g. by relying on efference-related information) as opposed to conditions
in which this modulation is transformation-specific. Otherwise, the resulting theory, by simply encompassing
two alternative hypotheses, would loose the predictive power (and empirical refutability) of the original
theories.
In conclusion, there is relevant empirical support to the thesis that several perceptual capabilities are mod-
ulated by action-dependent information, as predicted by defendants of radical theories of action/perception
relations. We suggested that further research directions will need to provide more constrained hypotheses
on what it means for perceptual information to be action-dependent. This requires in particular accounting
for conditions in which action modulates perception in a motor-specific way as opposed to conditions in which
this modulation is transformation-specific.

Acknowledgments

The ideas discussed in this paper have been presented at different stages of development over the last years.
We wish to thank David Philipona, Nivedita Gangopadhyay, Roberto Casati, as well as participants in the
Enactive Workshop (Paris, 2005), the Perception, Action and Consciousness Conference (Bristol, 2007), the
Interactivist Summer Institute (Paris, 2007), the 33rd SPP Annual Conference (Toronto, 2007) for valuable
feedback on earlier versions of this paper. Dario Taraborelli was partly supported by a grant from the ENAC-
TIVE Network of Excellence (IST-2002-002114). Both authors contributed equally to this work.

References

Bach y Rita, P. (1972). Brain mechanisms in sensory substitution. New York: Academic Press.
Barfield, W., & Furness, T. A. (1995). Virtual environments and advanced interface design. Oxford: Oxford University Press.
Berthoz, A. (2000). The brain’s sense of movement. Cambridge: Harvard University Press.
 M. Mossio, D. Taraborelli / Consciousness and Cognition 17 (2008) 1324–1340 1339

Blakemore, S. J., Frith, C. D., & Wolpert, D. M. (1999). Perceptual modulation of self-produced stimuli: The role of spatio-temporal
prediction. Journal of Cognitive Neuroscience, 11(5), 551–559.
Broackes, J. (2001). Experience, attention, and mental representation. BBS, 224(5), 978–979.
Carello, C., & Turvey, M. T. (2000). Rotational dynamics and dynamic touch. In M. Heller (Ed.), Touch, representation, and blindness
(pp. 27–66). Oxford, England: Oxford University Press.
Carello, C., Kinsella-Shaw, J., Amazeen, E., & Turvey, M. T. (2006). Peripheral neuropathy and object length perception by effortful
(dynamic) touch: A case study. Neuroscience Letters, 405, 159–163.
Cutting, J. E. (1986). Perception with an eye for motion. Cambridge, MA: MIT Press.
Duhamel, J. R., Colby, C. L., & Goldberg, M. E. (1992). The updating of the representation of visual space in parietal cortex by intended
eye movements. Science, 255(5040), 90–92.
Findlay, J. M., & Gilchrist, I. D. (2003). Active vision. The psychology of looking and seeing. Oxford: Oxford University Press.
Fodor, J. A., & Pylyshyn, Z. W. (1981). How direct is visual perception? Some reflections on Gibson’s ecological approach. Cognition, 9,
139–196.
Gallistel, C. R. (1980). The organisation of action: A new synthesis. Hillsdale, NJ: Erlbaum.
Gibson, J. J. (1950). The perception of the visual world. Westport, Connecticut: Greenwood Press.
Gibson, J. J. (1959). Perception as a function of stimulation. In S. Koch (Ed.). Psychology. A study of a science (Vol. 1, pp. 456–501). New
York: McGraw-Hill.
Gibson, J. J. (1960). Pictures, perspective, and perception. Daedalus, 89, 216–227.
Gibson, J. J. (1966). The senses considered as perceptual systems. Boston: Houghton Mifflin.
Gibson, J. J. (1979). The ecological approach to visual perception. Boston: Houghton Mifflin.
Gonzalez, J. C., & Bach-y-Rita, P. (2003). Perceptual adaptive recalibration: Tactile sensory substitution in blind subjects. Behavior and
Philosophy.
Held, R. (1965). Plasticity in sensory-motor systems. Scientific American, 213(5), 84–94.
Helmholtz, H. (1878/1977). Epistemological writings. Dordrecht/Boston: D. Reidel Publishing Co..
Hurley, S. (1998). Consciousness in action. Cambridge: Harvard University Press.
Hurley, S. (2001). Perception and action: Alternative views. Synthese, 291, 3–40.
Jacob, P. (2006). Why visual experience is likely to resist being enacted. Psyche, 12(1). Available from http://psyche.cs.monash.edu.au/
symposia/noe/Jacob.pdf.
Jacob, P., & Jeannerod, M. (2003). Ways of seeing: The scope and limits of visual cognition. Oxford: Oxford University Press.
Jarvilehto, T. (1999). The theory of the organism-environment system: III. Role of efferent influences on receptors in the formation of
knowledge. Integrative Physiological and Behavioural Science, 34, 90–100.
Koenderink, J. J., & Van Doorn, A. J. (1975). Invariant properties of the motion parallax field due to the movement of rigid bodies relative
to an observer. Optica Acta, 22, 773–791.
Koenderink, J. J. (1986). Optic flow. Vision Research, 26(1), 161–179.
Lee, D. N. (1976). A theory of visual control of braking based on information about time-to-collision. Perception, 5, 437–459.
Lee, D. N. (1980). Visuo-motor coordination in space–time. In G. E. Stelmach & J. Requin (Eds.), Tutorials in motor behavior
(pp. 281–295). Amsterdam: North-Holland.
Lenay, C., Cannu, S., & Villon, P. (1997).Technology and perception: The contribution of sensory substitution systems. Second
international conference on cognitive technology (pp. 44–53). Aizu, Japan, Los Alamitos: IEEE.
Lenay, C., Gapenne, O., Hanneton, S., Marque, C., & Genouëlle, C. (2003). Sensory substitution: Limits and perspectives. In Y. Hatwell,
A. Streri, & E. Gentaz (Eds.), Touch for knowing (pp. 275–292). Amsterdam: John Benjamins.
MacKay, D. (1962). Theoretical models of space perception. In C. A. Muses (Ed.), Aspects of the theory of artificial intelligence. New
York: Plenum Press.
MacKay, D. (1987). The organization of perception and action: A theory for language and other cognitive skills. New York: Springer.
Maturana, H. R., & Varela, F. J. (1980). Autopoiesis and cognition: The realization of the living. Boston studies in the philosophy of
sciences (Vol. 42). Dordrecht/Boston: D. Riedel Publishing Co.
Michaels, C. F., & Carello, C. (1981). Direct perception. Englewood Cliffs, New York: Prentice-Hall.
Milner, A. D., & Goodale, M. A. (1995). The visual brain in action. Oxford: Oxford University Press.
Nawrot, M. (2003). Depth from motion parallax scales with eye movement gain. Journal of Vision, 3(11), 841–851.
Noë, A. (2002). On what we see. Pacific Philosophical Quarterly, 83, 57–80.
Noë, A. (2004). Action in perception. Cambridge, MA: MIT Press.
Ono, M. E., Rivest, J., & Ono, H. (1986). Depth perception as a function of motion parallax and absolute-distance information. Journal of
Experimental Psychology: Human Perception and Performance, 12, 331–337.
O’Regan, J. K., & Noë, A. (2001). A sensorimotor account of vision and visual consciousness. Behavioral and Brain Sciences, 24(5),
939–1011.
Philipona, D., O’Regan, J. K., & Nadal, J.-P. (2003). Is there something out there? Inferring space from sensorimotor dependencies.
Neural Computation, 15(9).
Philipona, D., O’Regan, J. K., Nadal, J.-P., & Coenen, O. J.-M. D. (2004). Perception of the structure of the physical world using
unknown multimodal sensors and effectors. Advances in Neural Information Processing Systems, 15.
Philipona, D., & O’Regan, J. K. (2006). Color naming, unique hues and hue cancellation predicted from singularities in reflection
properties. Visual Neuroscience, 23, 331–339.
Poincaré, H. (1902). La science et l’hypothèse. Paris: Flammarion.
1340 M. Mossio, D. Taraborelli / Consciousness and Cognition 17 (2008) 1324–1340

Port, R. F., & van Gelder, T. (1995). Mind as motion. Cambridge, MA: MIT Press.
Prinz, J. (2006). Putting the Brakes on Enactive Perception. Psyche, 12 (1). Available from http://psyche.cs.monash.edu.au/symposia/noe/
Prinz.pdf.
Pylyshyn, Z. W. (2001). Seeing, acting, and knowing. Behavioral and Brain Sciences, 24(5), 999.
Reed, E. S. (1982). An outline of a theory of action systems. Journal of Motor Behavior, 14, 98–134.
Reed, E. S. (1986). Motor variability but functional specificity: Demise of the concept of motor commands. Behavioral and Brain Sciences,
9, 620–622.
Reed, E. S. (1996). Encountering the world: Toward an ecological psychology. New York: Oxford University Press.
Rogers, B., & Graham, M. (1979). Motion parallax as an independent cue for depth perception. Perception, 8, 125–134.
Rogers, B., & Graham, M. (1982). Similarities between motion parallax and stereopsis in human depth perception. Vision Research, 22,
261–270.
Sampaio, E. (1995). Les substitutions sensorielles adaptées aux déficits visuels importants. In A. B. Safran, & A. Assimacopoulos (Eds.),
Le déficit visuel. Des fondements neurophysiologiques à la pratique de la réadaptation. Paris: Masson.
Shaw, R. E., Kugler, P. N., & Kinsella-Shaw, J. M. (1990). Reciprocities of intentional systems. In R. Warren, & A. H. Wertheim (Eds.),
Perception and control of self-motion (pp. 579–619). Hillsdale, NJ: Erlbaum.
Sheerer, E. (1984). Motor theories of cognitive structure: A historical review. In W. Prinz & A. F. Sanders (Eds.), Cognition and motor
processes. Berlin: Springer Verlag.
Scholl, B. J., & Simons, D. (2001). Change blindness, Gibson and the sensorimotor theory of vision. Behavioral and Brain Sciences, 24(5),
1004–1005.
Stark, L., & Bridgeman, B. (1983). Role of corollary discharge in space constancy. Perception and Psychophysics, 34(4), 371–380.
Steinbach, M. J., Ono, H., & Wolf, M. (1991). Motion parallax judgments of depth as a function of the direction and type of head
movement. Canadian Journal of Psychology, 45, 92–98.
Stoffregen, T. A., & Bardy, B. G. (2001). On specification and the senses. Behavioral and Brain Sciences, 24, 195–213.
Stoffregen, T. A., & Bardy, B. G. (2004). Multisensory enactive interfaces and the global array, Online Paper, http://www.interdisciplines.org/
enaction/papers/3.
Stoffregen, T. A., Bardy, B. G., & Mantel, B. (2006). Affordances in the design of enactive systems. Journal of Virtual Reality Research, 10,
4–10.
Telford, L., Howard, I. P., & Ohmi, M. (1995). Heading judgments during active and passive self-motion. Experimental Brain Research,
104, 502–510.
Thelen, E., & Smith, L. B. (1994). A dynamic systems approach to the development of cognition and action. Cambridge, Mass: MIT Press.
Turvey, M. T. (1996). Dynamic touch. American Psychologist, 51(11), 1134–1152.
Turvey, M. T., & Carello, C. (1986). The ecological approach to perceiving-acting: A pictorial essay. Acta Psychologica, 63, 133–155.
Ujike, H., & Ono, H. (2001). Depth thresholds of motion parallax as a function of head movement velocity. Vision Research, 41,
2835–2843.
Van Boxtel, J. J. A., Wexler, M., & Droulez, J. (2003). Perception of plane orientation from self-generated and passively observed optic
flow. Journal of Vision, 3, 318–332.
Varela, F. (1979). Principles of biological autonomy. New York: North Holland.
Varela, F., Thompson, E., & Rosch, E. (1991). The embodied mind: Cognitive science and human experience. Cambridge, MA: MIT Press.
Wallach, H., & O’Connell, D. N. (1953). The kinetic depth effect. Journal of Experimental Psychology, 45, 205–217.
Wallach, H., Stanton, L., & Becker, D. (1974). The compensation for movement-produced changes in object orientation. Perception &
Psychophysics, 15, 339–343.
Warren, W. H. (1988). Action modes and laws of control for the visual guidance of action. In O. Meijer, & K. Roth (Eds.), Movement
behavior: The motor-action controversy (pp. 339–380). Amsterdam: North-Holland.
Warren, W. H. (1998). Visually controlled locomotion: 40 years later. Ecological Psychology, 10, 177–219.
Warren, W. H. (2006a). Enactive knowledge: Sensory-motor expectancies or perception-action invariants? Proceedings of the 3rd
International Conference on Enactive Interfaces, Montpellier, France, p. 15.
Warren, W. H. (2006b). The dynamics of perception and action. Psychological Review, 113(2), 358–389.
Wexler, M., Panerai, F., Lamouret, I., & Droulez, J. (2001a). Self-motion and the perception of stationary objects. Nature, 409, 85–88.
Wexler, M., Lamouret, I., & Droulez, J. (2001b). The stationarity hypothesis: An allocentric criterion in visual perception. Vision
Research, 41, 3023–3037.
Wexler, M. (2003). Allocentric perception of space and voluntary head movement. Psychological Science, 14, 340–346.
Wexler, M., & Van Boxtel, J. J. A. (2005). Depth perception by the active observer. Trends in Cognitive Sciences, 9, 431–438.
White, B. W., Saunders, F. A., Scadden, L., Bach y Rita, P., & Collins, C. C. (1970). Seeing with the skin. Perception and Psychophysics, 7,
23–27.
Withagen, R., & Michaels, C. F. (2005). On ecological conceptualizations of perceptual systems and action systems. Theory & Psychology,
15(5), 603–620.
Wolff, P. (2004). Position of code and code for position: From isomorphism to a sensorimotor account of space perception. Visual
Cognition, 11, 137–160.
Wolpert, D. M., & Flanagan, J. R. (2001). Motor prediction. Current Biology, 11, 729–732.