Available via license: CC BY 4.0
Content may be subject to copyright.
PAPER IN GENERAL PHILOSOPHY OF SCIENCE
European Journal for Philosophy of Science (2024) 14:35
https://doi.org/10.1007/s13194-024-00596-3
Abstract
Dierent species of realism have been proposed in the scientic and philosophical
literature. Two of these species are direct realism and causal pattern realism. Direct
realism is a form of perceptual realism proposed by ecological psychologists within
cognitive science. Causal pattern realism has been proposed within the philosophy
of model-based science. Both species are able to accommodate some of the main
tenets and motivations of instrumentalism. The main aim of this paper is to ex-
plore the conceptual moves that make both direct realism and causal pattern real-
ism tenable realist positions able to accommodate an instrumentalist stance. Such
conceptual moves are (i) the rejection of veritism and (ii) the re-structuring of the
phenomena of interest. We will then show that these conceptual moves are instances
of the ones of a common realist genus we name pragmatist realism.
Keywords Philosophy of science · Philosophy of cognitive science · Realism ·
Instrumentalism · Direct realism · Causal patterns
1 Introduction
If you were to go to the street and ask someone whether they think what they see
is real, you will likely get an armative answer. Yes, this is a real tree. Yes, that
approaching car is real. Yes, the building I am getting into is a real building. Some-
one might say something like “no, this is not real, we live in the Matrix”, but even
that person is probably joking and would move from the path of the approaching
Received: 29 April 2023 / Accepted: 2 July 2024
© The Author(s) 2024
Two species of realism
VicenteRaja1,2 · GuilhermeSanches de Oliveira3
Vicente Raja
vicente.raja@um.es; vicendio@gmail.com
1 Department of Philosophy, Universidad de Murcia, Murcia, Spain
2 Rotman Institute of Philosophy, Western University, London, Canada
3 Department of Psychology & Ergonomics, Technische Universität Berlin, Berlin, Germany
1 3
European Journal for Philosophy of Science (2024) 14:35
car to avoid being hit. Similarly, many people would claim the objects, processes,
and mechanisms that feature in our scientic theories and models are real objects,
processes, and mechanisms one can nd in the world. Black holes are out there,
with their singularity, their event horizon, and their ergosphere. The same goes for
action potentials in neurons, with their voltage-gated ion channels, or for the decod-
ing activities that produce proteins out of the genetic information in the DNA. Real-
ism about perception and science—i.e., the idea that both our perception and our
scientic theories and models oer a true characterization of the world—is just the
standard position in our everyday lives.
Realism is not, however, as dominant in academic settings. Many philosophers
and scientists defend dierent forms of nuanced realism or anti-realism regarding
both perception and scientic theories and models. In the cognitive sciences, for
instance, realist theories of perception are scarce. The most common position on
perception is that it involves the construction of an internal model (i.e., a representa-
tion) of the external environment that may or may not reect its real features (Shea,
2018). The challenge to perceptual realism entailed by this position goes all the way
from ctionalism concerning the phenomenal details of perceptual experience—e.g.,
the colors we perceive are not out there in the environment—to the characterization
of the whole perceptual experience as a kind of hallucination, as it often occurs in
the predictive processing literature (e.g., Clark, 2015; Seth, 2021). Also, perceptual
illusions are used to support the claim that perceptual experience is the product of a
process of mental construction based on belief and expectations and that, therefore, it
does not reect the real world as it is out there but only as we perceive it (e.g., Smith,
2002). The dominant paradigm of the contemporary cognitive sciences is thus gener-
ally anti-realist regarding perception.
The situation in the literature on scientic realism is not as extreme as in the case
of perceptual realism. With regard to scientic theories and models, realism is still
the go-to position for the majority of scientists and philosophers although such real-
ism is usually nuanced. It is not just naive realism. It is a form of scientic realism
that acknowledges scientic theories and models usually contain abstractions, c-
tions, idealizations, etc., that make them unlike mirrors reecting the real world. Sci-
entic theories and models are just approximately true about the target phenomena
they aim to explain. In this context, some of the objects, processes, and mechanisms
featured in scientic theories and models are approximately representing real objects,
processes, and mechanisms of the world. Contrary to this form of nuanced scientic
realism, an anti-realist position, instrumentalism, can also be found in the literature.
According to the proponents of instrumentalism, scientic theories and models are
mere instruments scientists use to achieve whichever their research goals are but do
not need to (or aim to) be true of the objects, processes, and mechanisms of the world.
An example of the quarrel between nuanced scientic realism and instrumentalism
may be found in contemporary discussions within the philosophy of biology and the
cognitive sciences. Bayesian and predictive models used in theoretical biology and
theoretical neuroscience, for instance, are interpreted both as mere scientic instru-
ments agnostic of ontological commitments (Andrews, 2021; van Es, 2020) and as
providing true descriptions of the components and processes of metabolic and cogni-
1 3
35 Page 2 of 24
European Journal for Philosophy of Science…
tive activities of living organisms (Kirchho et al., 2022). Both positions are indeed
still a matter of discussion within the philosophy of the life sciences.
Both when it comes to perception and to scientic practice, the general debate
between realism and anti-realism (e.g., ctionalism, instrumentalism) is based on
the opposition between two relatively simple observations relating to success and
failure. On the one hand, we seem to deal very well with our shared environment
on a daily basis. This fact points to the success of our perceptual capacities to cap-
ture the real properties of the world. Similarly, we are very successful in predict-
ing and explaining dierent aspects of the world when using our scientic theories
and models. Again, this fact seems to entail these theories and models capture real
objects, processes, and mechanisms. On the other hand, perception is not always
accurate and, more importantly, seems to vary from subject to subject in non-trivial
ways. For instance, we sometimes misjudge distances and shapes, seem to be dif-
ferently aected by illusions, and often perceive colors in dierent ways even when
no pathology is involved. Additionally, we can suer from perceptual hallucinations
under dierent conditions. All these events seem to support the idea of perception
as the mental construction of a model that is not always veridical of the real world.
Similarly, the history of science shows that scientic theories and models that at some
point in time are considered to capture the real objects, processes, and mechanisms of
the world are systematically substituted by more successful theories that posit dier-
ent sets of objects, processes, and mechanisms. In these cases, the substituted theory
comes to be considered no longer as veridical as it was once supposed to be, and it’s
abandoned in favor of a seemingly more veridical new one. In addition to this fact,
it is not dicult to nd dierent models of the same target phenomenon that, even
at the same moment of history, enjoy equivalent predictive and explanatory success
despite positing dierent objects, processes, and mechanisms. Together, these obser-
vations suggest that scientic theories and models do not represent the real world but
are mere instruments used by dierent scientists to pursue their dierent scientic
aims at dierent times or even the same moment in history. The debate is therefore
framed as a quarrel between acknowledging the success of perception and scientic
theories and models—and thus endorsing some form of realism—or acknowledg-
ing the epistemic fallibility and variability of perception and scientic theories and
models—and thus endorsing some form of anti-realism (e.g., ctionalism or instru-
mentalism). Such is the seemingly unsolvable dichotomy. Epistemic success or fal-
libility. Stability or variability. This paper stands against these dichotomies. Despite
their pervasiveness in the literature both on perception and on scientic modeling, we
contend it is possible to reconcile them. In other words, it is possible to accept and
to understand both success and fallibility, both stability and variability. They are not
opposed poles. One is not the negation of the other. It is possible to be a realist, and
therefore acknowledge that we deal with our environment quite well on a daily basis
and that our scientic theories capture reality, while accepting the fallibility and vari-
ability of our perception and scientic theories and models. Pragmatist realists make
the conceptual moves needed for this, and both direct realism and causal patterns
realism are instances that illustrate those moves. They are species of the pragmatist
realism genus.
1 3
Page 3 of 24 35
European Journal for Philosophy of Science (2024) 14:35
Our main aim in the following sections is to show that, although naive realism is
tout court incompatible with all anti-realist positions, the core conceptual moves of
these species of realism allow for fully accommodating the observations that classi-
cally led to ctionalism or instrumentalism. Thus, accepting them would eectively
resolve the tension between realism and anti-realism regarding both perception and
scientic theories and models. As just noted, one kind of realism able to do so is the
direct realism favored by ecological psychology in the context of perception (Gibson,
1966, 1979). Direct realism oers a general realist framework that takes perception
to deal with real properties and events of the environment while making room for dif-
ferent, and even incompatible, perceptual experiences within the same situation. This
is possible thanks to two of the basic conceptual moves of ecological psychology: (a)
the substitution of the notion of perceptual content by the notion of lawful specica-
tion and (b) the distinction between perceptual states and perceptual judgments. In
the case of scientic theories and models, a kind of realism able to accommodate
anti-realist observations is known as causal pattern realism (Potochnik, 2023a, b).
This kind of realism stems from the recognition of instrumentalism as a tenable posi-
tion and, more concretely, from acknowledging the crucial role of idealization in
scientic modeling. As in the case of direct realism, causal pattern realism is possible
due to two fundamental conceptual moves: (c) the rejection of veritism as the central
aim of science and (d) the distinction between the target phenomenon and the object
of knowledge in the context of scientic modeling.
In the rest of the paper, we evaluate the four conceptual moves needed to endorse
direct realism (Sect. 2), on the one hand, and causal pattern realism (Sect. 3), on the
other. As will become evident, the two pairs of conceptual moves respectively attrib-
uted to direct realism and causal pattern realism can be interpreted as two instances
of the same conceptual pair. The rst component of this pair is the rejection of truth
as the central foundation for realism. The second one is the distinction between phe-
nomena as we talk about them and the dierent kinds of structures/patterns embodied
by those phenomena. Given this, Sect. 4 will explore and argue in favor of pragmatist
realism. The conceptual moves provided by pragmatist realism are the reason why
we are in the position to reject the dichotomy between realism and instrumentalism.
To be clear, our argument is not aimed at convincing anti-realists in either domain (or
both) to become realists in either (or both). Rather, our goal is to articulate the rela-
tion between the realism/anti-realism debate in the two domains (a relation that so far
has remained largely unacknowledged), and, by clarifying the nature of what we see
as promising versions of realism in both, to propose a general framework that should
motivate those with realist sensibilities concerning either perception or science to
nd allies and support within the other domain. In this sense, and following a prag-
matist inspiration, the “cash value” of this work is not so much to oer a (new) form
of realism opposed or complementary to prominent realist claims in the philosophical
literature (e.g., Chang, 2022; Massimi, 2022; Rice, 2021) but to identify the concep-
tual moves—or philosophical maneuvers, if you wish—that make these two current
realist positions actually be realist. Once these conceptual moves are identied and
the general pragmatist strategy that underlies them is made explicit, we will have set
the conditions from which we can explore connections between dierent explanatory
practices and from which we can compare the two species of pragmatist realism with
1 3
35 Page 4 of 24
European Journal for Philosophy of Science…
other realist proposals in order to nd analogies and disanalogies, compatibilities and
dierences, shared or not shared philosophical maneuvers, and so on. But that is a
dierent story. In this work, we focus on the conceptual moves of direct realism and
causal pattern realism.1
2 Perceptual realism: ecological psychology
Ecological psychology is an approach to perception, action, and cognition that has
been developed during the past ve decades in the periphery of the dominant, infor-
mation-processing paradigm in the cognitive sciences. Ecological psychology has its
origins in the mid-20th century with the works of James and Eleanor Gibson and has
generated a consistent output of empirical and theoretical results since its inception.
In recent years, it has been regarded as one of the theoretical foundations of the 4E
approach to the cognitive sciences, joining eorts with enactivism and other theoreti-
cal paradigms in order to provide an alternative to the information-processing one
(Chemero, 2009; Di Paolo et al., 2017). Most important for the aims of our paper is
that ecological psychologists defend direct realism, which is the most prominent real-
ist take on perception we can nd in contemporary cognitive science.2
In a philosophical paper, Gibson (1967a) explicitly embraces direct realism in
perception as the corollary of the ecological approach. Put simply, according to Gib-
son and ecological psychologists afterwards, perceptual information lies in the struc-
ture of the ambient energy arrays (e.g., light, air, etc.) that surround organisms. This
structure is lawfully related to the layout of the environment these organisms inhabit
and, thus, perceptual information is lawfully related to real environmental states.
As perception is a function of perceptual information, it is lawfully related to those
real environmental states. Thus, if the ecological view of perceptual information is
accepted, perceptual realism is granted. In order to arrive at this conclusion, Gibson
makes two conceptual moves that would become the core of ecological psychology
and, consequently, the core of direct realism. We turn to them now.
2.1 Conceptual move A: specication instead of content
In the theoretical context of ecological psychology, the notion of specication is the
foundation of the lawful relationship between environmental states and perceptual
information (see Segundo-Ortin et al., 2019). The rst step to explain specication is
to lay out the big picture of the ecological approach to perception. Its central tenet is
that perception is direct (Chemero, 2009; Michaels & Carello, 1981; Turvey, 2018).
1 We thank an anonymous reviewer for the suggestion of making explicit the “cash value” of this piece
of work.
2 Notice that the notion of direct realism we defend here is the one that comes directly from the ecological
approach to perception and action in experimental psychology (Gibson, 1967a). We are aware notions of
direct and indirect realism are also used in the eld of analytical philosophy of mind. There might be analo-
gies and disanalogies between the ecological notion of direct realism and the one used in the philosophy
of mind. We will not explore these issues in this paper. All references to direct realism are references to
the ecological notion.
1 3
Page 5 of 24 35
European Journal for Philosophy of Science (2024) 14:35
This means, contra the information-processing paradigm, that perception is not the
outcome of a process of enrichment and disambiguation of sensory stimulation. In
other words, perception is not about constructing an internal model of the exter-
nal environment by patching poor sensory stimulation with internal resources (e.g.,
memories, priors) of non-sensory, non-perceptual origin. Perception is not mediated
by the construction of a mental representation of the external environment. In this
sense, the case for anti-realism in perception weakens. The mediatory processes that
make use of memories, inferences, expectations, priors, etc., and fuel anti-realism
are just absent within the theoretical context of ecological psychology. But how is
perception possible without all these processes? Here’s where the notion of specica-
tion does its job.
Regarding perception, the alleged need for mediatory processes to build up an
internal model of the environment rests on one single assumption: that whatever
information we get through our senses is just not enough to support perception (and,
by extension, intelligent behavior). This is often called the poverty of stimulus argu-
ment and was rst attributed to Noam Chomsky in his critique of B. F. Skinner’s
Verbal Behavior (see Chemero, 2009). The argument goes like this: at any single
point in time sensory stimulation is just too scarce and too ambiguous to be the sole
basis of perception; thus, mediatory processes must construct an internal model of the
environment that compensates for such scarcity and ambiguity of sensory stimula-
tion. Ecological psychology rejects this argument. The works of the Gibsons and later
ecological psychologists have consistently shown that sensory stimulation carries
enough information for perception when properly described (see, e.g., Gibson & Gib-
son, 1955; Gibson, 1966, 1979; Lee, 2009; Turvey et al., 1981; Warren, 2021). The
proper description of sensory stimulation in terms of perceptual information involves
an ecological understanding of the ambient energy arrays surrounding organisms in
a given environment.
One of these ambient energy arrays is, for instance, the ambient optic array. The
ecological understanding of the ambient optic array is what Gibson named ecological
optics and focuses on the structure of light in that array (Gibson, 1961; Tsao & Tsao,
2021). This structure depends on the sources of illumination (e.g., sun, lamps) and the
surface layout of the environment: the dierence in illumination (i.e., the structure of
light) over and under my desktop depends on the position of the lamps in my oce,
the height of the desktop, the properties of the surfaces of the walls and the oor of
the oce, etc. Due to this dependence and the laws of optics, the structure of light
in the ambient optic array is lawfully related with the layout of the environment.
Therefore, at least some properties of this structure are specic to the properties of
the environment: this specicity means that coming into contact with the structured
light in this case is unambiguously and unequivocally informative, for the perceiving
organism, about its environment.
Additionally, when organisms move around the environment, the structure of the
ambient optic array available to them changes with their movement. This change
is known as optic ow (Warren, 1998; Matthis et al., 2022). For the same reasons
detailed in the case of ecological optics, some properties of optic ow are specic to
the properties of the environment and the movement of the organisms in it: accord-
ingly, patterns of change in optic ow are informative for the organism not only
1 3
35 Page 6 of 24
European Journal for Philosophy of Science…
about the environment in general, but precisely about the nature of the organism’s
relation to it. Concretely, ecological psychologists claim that the properties of the
optic ow that remain invariant under dierent transformations specify the layout
of the environment, while some of the changes in the ow specify the movements of
the organism. For example, when an organism engages in forward locomotion while
looking ahead, the focus of expansion of its optic ow invariably species the place
in the environment it is heading to, while the centrifugal expansion of the global optic
ow from the focus of expansion species the locomotive movement itself; walking
backwards, in contrast, generates a centripetal pattern in optic ow that is specic to
(or lawfully related to) this way of moving in the environment. Within the ecological
literature, perception is then dened not as, say, the internal representing and process-
ing of information about the environment, but rather as the detection of the properties
of the ambient energy arrays that specify the real properties of both the environment
and the organism-environment interactions.
Ecological psychologists have provided a good amount of experimental support
to the idea of specication and to the idea of perception as a function of specic per-
ceptual information. We refer the reader to the ecological literature for more details
on these issues as we will not further defend the ecological take on perception here
(see, e.g., Gibson, 1966, 1979; Lee, 2009; Turvey et al., 1981; Warren, 2006, 2021;
and references therein). We want, however, to point out one of the main conceptual
moves as a consequence of accepting the ecological story regarding perception. As
we have already noted, the information-processing paradigm typical of the cognitive
sciences takes perception to involve the construction of an internal model of the envi-
ronment. Such a process of construction is indeed the root of mainstream anti-realism
(ctionalism, etc.) with regard to perception. And this is so because the constructed
internal model can be true or false of the environment depending on the dierent ele-
ments from which it has been constructed. In other words, the content of the internal
model may represent or misrepresent the real world and, in any case, it will never
constitute a direct reection of the properties of that world. It will always involve
some degree of ction as a consequence of its constructive nature.
An intuitive “realist” alternative would be to downplay the threat of misrepre-
sentation and to hold, for instance, that the content of our internal models are (virtu-
ally) always accurate or true enough, thus guaranteeing that our perceptual states are
reliable and trustworthy. But this is not the ecological version of direct realism. The
ecological approach to perception, in contrast, does not relate perceptual states to
environmental states in terms of truth or content (Segundo-Ortin et al., 2019). The
relationship between the structure of the ambient energy arrays and the environment
is not characterized in terms of truthfulness or falsehood, but in terms of lawfulness:
a given property of the optic ow, for instance, is not true or false of the environ-
ment or the organism, but it just lawfully emerges from the organism-environment
interactions. Similarly, perceptual states are functions of perceptual information but,
crucially, they are not true or false of that perceptual information. According to eco-
logical psychologists, perceivers detect ecological information by resonating or being
attuned to it (Raja, 2018, 2019, 2021). They are more like a radio or a scanner tuning
to the right signal than a device constructing a model of that signal. Attunement (or
resonance) is obviously not cast in terms of truth-values: radios tune or not, but they
1 3
Page 7 of 24 35
European Journal for Philosophy of Science (2024) 14:35
are not true or false of a given radio station. Thus, ecological psychology eectively
removes the notion of truth from the forefront of the theory of perception. While
“naive realism” might be construed as a belief in the veridicality of perception, direct
realism is not at all related to veridicality, instead construing perception as something
that cannot have truth value—at least when it comes to traditional views of truth in
terms of correspondence, representational accuracy and so on. This does not mean
that the notion of truth must be completely abandoned, of course. In a deeply prag-
matist vein, truth is not abandoned but reconceived when accepting direct realism.
For particular contexts and descriptions, truth, correctness, and accuracy still apply
to activities based on and guided by perceptual knowledge (see Sect. 2.2 and Sect. 4;
see Raja & Chemero, 2020). They just do not apply to the concrete relationship of
fundamental perceptual states and the environment. Perception gives us knowledge
of some real properties of the environment without perceptual states being the sort of
thing that can be true or false. But how can it be?
2.2 Conceptual move B: separating perceptual states and perceptual judgments
Philosophy of perception and information-processing-based cognitive science share
what can be called a descriptive view of perceptual states. In the case of philosophy
of perception, perceptual states are often considered to be judgments about the envi-
ronments—aka propositional attitudes (e.g., Byrne, 2005; McDowell, 1994). Percep-
tual states are of the form “that apple is red” or “this car is approaching”. In other
words, perceptual states are a form of categorical judgment that identify a subject
of predication and predicate something about it. The content or meaning of these
states is of course evaluated in terms of truth and falsehood. The case of information-
processing-based cognitive science is similar. Although not always framed in terms
of propositional attitudes, perceptual states are taken to be representations of objects
and their properties (Hafri & Firestone, 2021; Shea, 2018). In this sense, perceptual
states also involve a process of categorization (i.e., identifying an object and predi-
cating something about it) and can be evaluated in terms of truth and falsehood as
they can represent or misrepresent the target environmental state.
Ecological psychology frontally rejects the characterization of perceptual states
as perceptual judgments or representations. As we have already noted, ecological
psychologists take perceptual states to be functions of perceptual information. Per-
ceivers detect perceptual information and that’s the way they meaningfully relate to
their environment. But, if perceptual states are not representations of the environment
or categorical judgments, in what way are they meaningful? How do they provide
cognitive access to the world? According to the ecological approach, perception is
of aordances and they are the foundation of perceptual meaning (Heras-Escribano,
2019). Aordances are opportunities for interaction that organisms nd in their envi-
ronments: the edibility of an apple, the walk-ability of the ground, the step-ability
of stairs, the grab-ability of a mug, etc. Perceptual information is specic to these
opportunities for interaction and detecting it is perceiving an aordance. Importantly,
perceiving an aordance is not based on a judgment about the existence of the aor-
dance. For instance, when a goat perceives the jump-ability of a cli, there is no
perceptual state of the form “this cli is jumpable” in the mind/brain of the goat.
1 3
35 Page 8 of 24
European Journal for Philosophy of Science…
Indeed, there is no category of “cli” in the goat to begin with. The goat meaning-
fully relates with its environment by a perceptual experience that—when properly
tuned, in the radio metaphor described above—allows for a proper control of action.
But that particular perceptual experience does not involve any kind of judgment or
categorization. Even though we might describe the target phenomenon as a cli being
jumpable or not, the perceptual experience is just the function of the structure of
light surrounding the goat and specic to a particular environmental state and the
goat’s relation to it. Notice that this does not mean that we do not make categorical
judgments or that we do not have categorical knowledge. To be clear, categoriza-
tion is something we—and maybe other animals—do in many sociocultural contexts.
And categorical judgments and categorical knowledge are perfectly compatible with
direct realism. Ecological psychologists take the perception of aordances to be the
fundamental perceptual process and, therefore, to be delivering the fundamental per-
ceptual knowledge. Afterwards, this knowledge can inform categorization practices.
For instance, aordances seem to inform linguistic categories and other forms of
representations, as the empirical evidence in ecological psychology and other elds
suggests (e.g., Wilford et al., 2022; Castellini et al., 2011; Snow & Culham, 2021).
However, perception of aordances and categorization are not analogous.3
In the general case, the disanalogy between the perception of aordances and per-
ceptual judgments or categorizations is explicitly discussed in a sidebar of Gibson’s
last book:
To perceive an aordance is not to classify an object. The fact that a stone is a
missile does not imply that it cannot be other things as well. It can be a paper-
weight, a bookend, a hammer, or a pendulum bob. It can be piled on another
rock to make a cairn or a stone wall. These aordances are all consistent with
one another. The dierences between them are not clear-cut, and the arbitrary
names by which they are called do not count for perception. If you know what
can be done with a graspable detached object, what it can be used for, you can
call it whatever you please… You do not have to classify and label things in
order to perceive what they aord. (1979, p. 134).
3 An anonymous reviewer suggests a possible problem: by not paying due attention to perceptual judg-
ments, direct realists are forced to abandon some of the traditional notions of perceptual knowledge. We
agree with the anonymous reviewer that judgments involving categorization, such as “the computer is on
the table”, are traditionally understood as a kind of perceptual knowledge. In the case of direct realism,
however, they are seen as distinct cognitive activities informed by perceptual knowledge and constrained
by other cognitive abilities, like learning, language, social skills, and so on. The key is that what direct real-
ists—as opposed to the traditional theories of perception—are willing to call “perceptual knowledge” does
not include non-perceptual abilities. Traditional theorists are willing to include non-perceptual resources
in perception and, therefore, are willing to accept some form of inference that goes from non-categorical
sensations to categorical judgments. An inference may involve memory, information-processing, construc-
tion of mental representations, ltering, edge detection, amodal completion, and so on. These processes
are precisely the ones that lead to anti-realism in perception, as perceptual knowledge is taken to be the
product of an event of subjective construction—or controlled hallucination, according to some (Seth,
2021). Direct realists do not accept such kind of inference in (at least the fundamental bits of) perceptual
knowledge, securing a fundamentally realist access to the environment and a safe, realist foundation for
further practices of categorization and other kinds of knowledge.
1 3
Page 9 of 24 35
European Journal for Philosophy of Science (2024) 14:35
The sidebar is illuminating for a fundamental reason: it frames aordances in oppo-
sition to xed categories. By doing so, it recognizes the variability of perceptual
experiences beyond our categorization practices. For instance, even when we classify
a target situation as involving some specic object, like a stone, that does not mean
that our perceptual experience is xed. Our description of the target situation is not
the object of experience. The objects of experience are aordances, and detecting
perceptual information is perceiving them.
This is the last component of direct realism. Aordances are both objects of per-
ceptual experience and real elements of the organism-environment interactions. The
lawful and specifying character of perceptual information is what ensures the reality
of perceived aordances without the need for judgments, categorization, or evalua-
tion in terms of truthfulness and falsehood. In this sense, direct realism is founded on
the identication of the object of perception beyond our descriptions of the environ-
ment and the rejection of the evaluation of perception in terms of its veridicality. At
the same time, as aordances are opportunities for interaction—and therefore con-
cern the environment and the perceiver at the same time—direct realism ensures that
similar situations (i.e., situations involving the same description) may provide dier-
ent aordances to dierent organisms or to the same organisms at dierent moments
in time. Thus, direct realism dissolves the tension between the stability emphasized
by realism and the variability emphasized by anti-realism.
3 Scientic realism: causal patterns
In philosophy of science, realism can be understood as an attitude toward the suc-
cess of science, namely the perspective that the success of science stems from the
truth of scientic theories. In particular, as it is often construed, the realist stance
holds that it is because theories are true (at least approximately) that they can be used
for generating successful predictions and explanations of phenomena in the world.
Given this broad characterization, realism is traditionally tied to discussions concern-
ing unobservable entities and processes posited by our theories: in contrast with the
(anti-realist) instrumentalist view that empirical success does not warrant conclu-
sions about truth, realists hold that successful theories are useful precisely because
they are in fact true, because they actually (even if not completely accurately) capture
what the world is like, including the entities and processes posited that are not subject
to direct observation. This realist stance is neatly captured in the claim that “mature
and genuinely successful scientic theories should be accepted as nearly true” (Psil-
los, 1999, p. xv).
In her recent work, Angela Potochnik has oered an elegant and nuanced alterna-
tive version of scientic realism, one that takes seriously the success of science and
realist intuitions about this success, but that reorients realist commitments so as to
make them compatible with what might otherwise seem to be an anti-realist stance
on scientic theories and models. In her 2017 book Idealization and the Aims of Sci-
ence, Potochnik emphasizes how idealization is “rampant and unchecked” in science:
scientists routinely build and use models that dier, sometimes even grossly, from the
phenomena of interest, not only simplifying and omitting detail, but also including in
1 3
35 Page 10 of 24
European Journal for Philosophy of Science…
the model features known to be absent in the real-world phenomena. And not only is
idealization widespread, but it’s typically embraced by scientists as unproblematic,
both as a general practice and as a characteristic of particular models. Dierent mod-
els that are seen as scientically successful can work under dierent assumptions,
they can model dierent aspects of the same target phenomenon, and they can even
be totally incompatible—and still all of this does not undermine scientic realism.
This is possible thanks to two key conceptual moves.
3.1 Conceptual move C: understanding rather than truth
Potochnik’s view can be interpreted as revealing how science succeeds through fail-
ure. As just described, models knowingly and often intentionally fail as representa-
tions of real-world systems and phenomena: models can be inaccurate, incomplete,
and even outright wrong about their target phenomena in various respects—this
means that, if judged in terms of their representational content, models knowingly
and often intentionally fail to give us the truth about the world. And, with the support
of examples from various cases of scientic practice, Potochnik further emphasizes
that many of the best, most successful scientic theories and models are evaluated
by scientists as being good and successful not despite idealization (i.e., a failure to
be true in various respects) but because of it: scientists often treat idealizations as
features that make models and theories more, rather than less, useful and success-
ful. When confronted with such a puzzling observation—scientic success arising
through failure with regard to truth—the philosopher has to decide whether to con-
clude that scientists are wrong or to conclude that philosophical conceptions about
science need to be revised. Potochnik takes the latter path.
The key idea is to reconsider what we take the aims of science to be. “Idealizations
cannot directly contribute to science’s epistemic success in virtue of their truth,” after
all by denition they are not true (2017, p. 94). Instead, as Potochnik provocatively
puts it, “science isn’t after the truth”: following Elgin (2004, 2017) and others, she
proposes that the fundamental epistemic achievement of science, and its main aim, is
understanding rather than representational accuracy or truth.
For this to work, she explains, “understanding” must mean more than just the
feeling of understanding. Sometimes, when confronted with a problem, people can
mistakenly feel they’ve understood the situation and move on, only to later nd out
(or be told by someone else) that they had not in fact correctly understood what was
going on. “Understanding” includes this psychological dimension, the felt experi-
ence that a tension has been resolved, a problematic situation has been addressed, and
insight has been gained. But for it to be genuine, understanding must also include an
epistemic dimension: you only achieve understanding when, in addition to feeling
like you have grasped something, you did in fact successfully grasp it.
This “dual nature” of understanding, as both psychological and epistemic, is cru-
cial in Potochnik’s account. Truth ticks one of the boxes: creating a true descrip-
tion constitutes an epistemic achievement, insofar as a true description succeeds in
capturing what the thing described is actually like. But truth does not guarantee the
psychological dimension of understanding, and it can even prevent it. A maximally
comprehensive and accurate description of some system would by denition be
1 3
Page 11 of 24 35
European Journal for Philosophy of Science (2024) 14:35
true. But sometimes less is more. Excessive detail can in many cases be a source of
confusion, with irrelevant truths getting in the way of intelligibility and insight by
occluding what actually matters. And this is because, in science, “what matters” is
not dened only in terms of the phenomena, but also of the people studying them.
Science is not in the business of compiling truths about the world: science is some-
thing that humans do, and what scientists are after is understanding. And as limited
beings grappling with a complex world, we have to simplify and idealize: “Science is
tailored to human needs and thus human limitations, which leads to a focus on rather
simple patterns that contribute to human understanding and inuence” (Potochnik,
2017, p. 57).
With its emphasis on truth, traditional realism seems out of touch with real-world
scientic practice, especially in model-based research, where “less accurate explana-
tions are sometimes better than more accurate alternatives” (Potochnik, 2023a, p.
154). But the problem here is the assumption that a realist attitude toward the suc-
cess of science must be committed to veritism, namely the assumption that truth is
the fundamental epistemic achievement. Shifting our emphasis to understanding (in
both its psychological and epistemic dimensions) allows us to embrace the role of
idealization and still remain realists: by contributing to understanding, “idealizations
contribute directly to the epistemic success of our scientic explanations” (2023a, p.
154). But how can this be?
3.2 Conceptual move D: separating target phenomena and causal patterns
Scientic realism, we have seen, is a positive attitude about the success of science,
and in particular, as traditionally construed, one in which science is seen as success-
ful because, and to the extent to which, it truthfully represents how the world works.
As an alternative, and to disentangle it from veritism, Potochnik oers the following
broad characterization of what realism ultimately boils down to: “realism is the idea
that our best scientic accounts qualify as epistemic achievements and yield knowl-
edge of the world” (2023a, p. 10, italics original). Given this characterization, and
given the shift to an emphasis on understanding that accommodates the contribution
of idealizations, the question that arises, then, concerns what we’re realists about, if
not about the truth of scientic theories, models, explanations and accounts of the
world. Potochnik explains:
I think we should be realists about—that is, posit that we have scientic knowl-
edge of—the objects of well-corroborated theoretical claims. But those objects
are not unobservable entities, mechanisms, or even the phenomena under
investigation. Rather, science’s theoretical claims (when successful) often yield
knowledge of causal patterns. (Potochnik, 2023b, p. 4)
The key move here is that of distinguishing target phenomena from causal patterns.
Put simply, causal patterns are regularities in dependence relations, that is, regulari-
ties in “how a shift in one thing changes another” (2017, p. 29). One and the same
system or phenomenon embodies multiple causal patterns, while one and the same
causal pattern can be, “perhaps with deviations and exceptions, embodied by some
1 3
35 Page 12 of 24
European Journal for Philosophy of Science…
limited range of phenomena” (2017, p. 95). Consider, for instance, the causal pat-
tern of predator-prey dynamics with their cycles of relative population uctuation.
This causal pattern was notoriously formalized in the Lotka-Volterra equations based
on observations of sh populations in the Adriatic Sea, but it is clearly not limited
to that particular context. The same causal pattern is also exhibited by many other
biological populations in other places throughout the world, whether aquatic (such
as phytoplankton and benthos in the San Francisco Bay: see, e.g., Dame & Prins,
1997) or terrestrial (say, polar bears and seals, or rabbits and foxes in their respective
niches), and it appears also in non-biological systems, such as in the US and British
economies (see, e.g., Goodwin, 1967, Desai, 1984, Mohun & Veneziani, 2016). The
same causal pattern is there to be found in many dierent real-world systems. At the
same time, however, any of these real-world systems exhibits a practically endless
number of dierent causal patterns beyond this particular one. In addition to the
causal patterns of predator-prey dynamics just mentioned, and more famously so in
philosophical circles, the San Francisco Bay also exhibits the fundamentally dierent
hydraulic causal patterns modeled in the U.S. Army Corps of Engineers Bay Model,
the scale model built in the 1950s for testing the viability of a project for building a
dam in the Bay area (Potochnik et al., 2018). And similarly with the British economy,
which besides embodying causal patterns characteristic of predator-prey relations,
also exhibits distinct patterns of causal dependency between rates of taxation, sav-
ings, and investment in health care and education, patterns that are well-known in the
philosophical literature for having been modeled with pipes and tanks in the Phillips
machine (ibid). Examples like these are everywhere to be found. One and the same
real-world system embodies a number of dierent causal patterns that can each be
studied in their own right, whereas one and the same causal pattern might be present
in a number of distinct and otherwise very dierent systems. The key idea at play
here, then, is that scientists pursue and often achieve understanding of target phe-
nomena, but this is not by means of aiming for, and succeeding in developing, true
(even approximately) accounts of the phenomena, but rather by gaining knowledge
of relevant causal patterns that the phenomena of interest embody.
The terms “relevant” and “of interest” in the previous sentence point to the psy-
chological dimension of this version of scientic realism and the view of science it’s
built on. As we have seen, understanding is inherently agential: unlike the absolute
truths of impersonal descriptions, it’s always someone who understands or doesn’t
understand something. Accordingly, given the multiplicity of causal patterns embod-
ied by phenomena, which causal pattern is relevant can vary with context and be
dierent for dierent scientists and scientic projects: “a causal pattern’s ability to
explain (or, equivalently, to engender understanding) depends on the research inter-
ests of those seeking explanation” (2023b, p. 176). But this is still a realist position
insofar as there is a fact of the matter concerning the existence (or not) of the causal
pattern of interest. Both the causal pattern of predator-prey relations and the dierent
causal patterns of water currents, sediment movement and so on really are present in
the San Francisco Bay, but only the latter were directly relevant for researchers trying
to answer the question “what would happen if we build a dam here?”. Information
about causal dependence and the scope of dependence is useful, illuminating and
explanatory (or conducive to understanding) because it’s “precisely the information
1 3
Page 13 of 24 35
European Journal for Philosophy of Science (2024) 14:35
needed to predict and intervene on our world: information about what factors inu-
ence a phenomenon, and in what circumstances” (Potochnik & Sanches de Oliveira,
2020, p. 1314). And causal patterns are regularities that we can have actual, “full-
edged” knowledge of, as well as fail to—what’s context dependent in scientic
knowledge about causal patterns is not whether the causal pattern is real, but whether
it is of interest, that is, “the cognitive value to the explainers or knowers” (Potochnik,
2023b, p. 176).
The last point to be articulated concerns how idealization enters the picture—that
is, how to link understanding of target phenomena through knowledge of causal pat-
terns, on the one hand, and the way model-based research violates veritist expecta-
tions, on the other. And the key, quite simply, is that even while grossly diering
from various aspects of the target phenomena, scientic models can contribute to
understanding by successfully capturing the relevant causal patterns. Just as a given
causal pattern can be present in dierent phenomena, so can models embody the same
causal pattern, thereby enabling interventions that support insight into the phenom-
enon of interest. By embodying the relevant causal patterns in each case, the scale
San Francisco Bay model, the Lotka-Volterra equations, and the Phillips machine all
made it possible for researchers to gain understanding about the phenomena under
investigation, despite the many obvious dierences between the models and the real-
world systems of interest.
The way in which models dier from target systems and phenomena is, of course,
traditionally construed as the way in which models misrepresent those targets. The
scale of the San Francisco Bay model is disproportional for width and depth com-
pared to the actual bay, whereas real-life preys don’t have unlimited access to food,
and the British economy does not run on water and is subject to ination, something
that the Phillips machine didn’t account for. These “sins of omission and commission”
(Callender & Cohen, 2006), or dierences due to what models neglect and what they
incorrectly posit, are typically described as ways in which models, as representations
of their targets, are false about those targets. And although Potochnik sometimes talks
about models along these lines, in representational terms (see, e.g., Potochnik, 2017),
her emphasis on real causal patterns embodied by phenomena and by models, on the
one hand, and her emphasis on understanding as the central epistemic aim of science,
on the other, are suggestive of the viability of an alternative interpretation in line with
recent non-representational accounts of models as tools, artifacts and instruments,
such as “radical artifactualism” (Sanches de Oliveira, 2022a; see also Sanches de
Oliveira et al., 2021). A common representationalist assumption is that the epistemic
worth or value of model-based research is best understood in terms of representa-
tional relations between models and target phenomena. In contrast with accounts that
adopt an artifactualist ontology (treating models as tools) but remain committed to
this representationalist epistemology (holding that epistemic success is tied to repre-
sentation), radical artifactualism proposes that the tool-like or instrumental nature of
models suces also to account for their epistemic value: a model can help “advance
scientic understanding of some real-world system by being similar to that system
in some action-relevant way,” such as “when model-artifacts enable manipulations
that are similar to manipulations of interest in some real-world system” (Sanches de
Oliveira, 2022a, p. 26). Arguably, the interactions that models enable can be under-
1 3
35 Page 14 of 24
European Journal for Philosophy of Science…
stood in terms of the causal patterns that models embody, some of which may be
relevant and useful in the context of research targeting dierent phenomena: because
the model embodies those causal patterns (e.g., literal water current and sediment
movement patterns in the scale Bay model or abstract cyclic relative uctuation pat-
terns in predator-prey models), interacting with the model can be a means to gaining
knowledge of the causal pattern and, accordingly, it can inspire similar interven-
tions in other systems that also embody the same causal pattern; yet recognizing
this does not necessitate ontologically or epistemologically interpreting the model
(say, the scale Bay model or the Lotka-Volterra equations) in representational terms
as a description or depiction of the target and its causal patterns, just as nding the
same causal pattern in dierent real-world systems (say, predator-prey dynamics in
biological populations A and B, or water current patterns in bays X and Y) doesn’t
entail treating one system as a representation of the other. In this perspective, the
so-called “target phenomena” aren’t understood as targets that the model is supposed
to represent, but rather as targets of investigation that provide the context in which
building and operating with tools (i.e., models) that embody certain causal patterns
can be epistemically useful because it leads to better understanding of those phenom-
ena through knowledge of their causal patterns. And similarity with regard to the
causal patterns that model and target both embody can be (and often are) exploited
for making claims which may be true or false about the target, but the similarity itself
has no truth value: similarity and dissimilarity between two things with regard to the
causal patterns they embody can explain why engaging with one sometimes leads to
a better understanding of the other, but the similarity and dissimilarity do not on their
own make one a (better or worse) representation of the other. If this is right, then, by
connecting to emerging views of models as tools while remaining neutral with regard
to claims concerning representation, it seems that scientic realism based on causal
patterns is even more attractive and generally applicable than Potochnik may have
anticipated.
To conclude, in the realism about causal patterns explored here, the same target
phenomenon can embody dierent causal patterns—not like objects embody Aris-
totelian forms, but more like time series embody trends or higher-order structures.
Dierent models used in investigations of the same phenomena can dier from one
another, and even be incompatible with each other, by targeting those dierent causal
patterns. In the end, the epistemic value of models is not properly construed in terms
of whether they are true or false about the target phenomena: rather, models contrib-
ute to the epistemic success of science because they capture, more or less well, causal
patterns embodied by the phenomena, and through this, enable better understanding
of those causal patterns. Which causal patterns of which phenomena are of interest
depend on the goals of dierent scientic projects. But the causal patterns are real,
both as aspects of the real-world phenomena and of the models. Causal pattern real-
ism is scientic realism without veritism.
1 3
Page 15 of 24 35
European Journal for Philosophy of Science (2024) 14:35
4 Two species of realism, one genus
In principle, direct realism and causal pattern realism are independent from each
other, and the elds where they apply—perception and science—are not necessarily
connected (cf. Sanches de Oliveira et al., 2023). However, the two forms of realism
share the same kind of conceptual moves, and, we propose, can be understood as
two species of the same realist genus. To begin articulating this relation and corre-
spondence between the two, a summary of the convergences discussed up to now is
provided in Table 1.
Direct realism and causal pattern realism coincide in moving away from the
emphasis on truth that, in each of their domains, tends to motivate the tension
between realist and anti-realist perspectives. In the domain of perception, direct real-
ism does this by abandoning the notion of perceptual content based on truth-values
in favor of the notion of specication based on natural law (conceptual move a). And
in the domain of science, causal pattern realism does this by rejecting veritism and,
instead, construing (or rather, recognizing in scientic practice) the more fundamen-
tal role of understanding rather than truth as the primary epistemic aim of scientists
(conceptual move c). In each case, the analogous move is made possible by the way
in which direct realism and causal pattern realism both reframe the object of interest
in their domains—another key respect in which the two coincide. For direct realism,
moving away from the typical emphasis on truth is possible because in ecological
psychology the object of interest is reframed by separating perceptual states from
Table 1 Convergences between direct realism and causal pattern realism that result from the analogous
conceptual moves at play in the two forms of realism as described in Sects. 2 and 3 (i.e., a in direct realism
alongside c in causal pattern realism, and b in direct realism alongside d in causal pattern realism)
Direct Realism (Perception) Causal Pattern Realism (Science)
Variability One and the same object has poten-
tially many dierent aordances, and
dierent aordances can be more or
less relevant for dierent people at
a given point in time as well as for
the same person at dierent points
in time.
One and the same model and one and the
same phenomenon embody potentially many
dierent causal patterns, and dierent causal
patterns can be more or less relevant for
dierent scientists (or scientic projects) at
the same time or for the same scientists (or
scientic project) at dierent points in time.
Stability Perceptual states specify, or are
lawfully related to, organism-envi-
ronment interaction, making them
reliable guides to action through the
detection of aordances.
Models that embody a causal pattern that
is also embodied by some target system or
phenomenon can be reliably used for inter-
ventions that generate understanding of that
causal pattern and, through this, understand-
ing of the target.
Epistemic suc-
cess without
veritism
Perceptual states can be used for
categorization and to form judgments
(which have truth value, that is, can
be evaluated as true or false about
aspects of the environment), but these
are secondary; the primary aim of
perception is the guidance of action
through the detection of aordances.
Models can be used to inform descrip-
tions of target phenomena, and they may
themselves be interpreted as representing
the phenomena more or less accurately in
some way or other (thus enabling evalua-
tion in terms of truth or falsehood), but this
is secondary; the primary aim of modeling
is to engender understanding by enabling
interventions that are relevant in the context
of investigation about “target” phenomena.
1 3
35 Page 16 of 24
European Journal for Philosophy of Science…
perceptual judgments (conceptual move b): in contrast with the common assump-
tion that judgments are the starting point for perception, direct realism reverses this
order, seeing judgments as enabled by, and potential products of, perceptual experi-
ence. And for causal pattern realism, moving away from the usual emphasis on truth
is possible because the object of interest is reframed through a separation between
causal patterns and target phenomena (conceptual move d): rather than assuming that
the accumulation of truths about target phenomena is the primary measure of epis-
temic success, this variety of realism recognizes the more fundamental importance
of causal patterns, and in particular, the goal of understanding the causal patterns
embodied by target phenomena through investigation of other objects that embody
the same causal patterns (i.e., models).
Through these analogous conceptual moves (i.e., a and c, b and d), direct real-
ism and causal pattern realism both reject the emphasis on truth that is common in
their respective contexts, yet this is not to say the two frameworks entirely abandon
any notion of truth. Both have a role for truth, it’s just not the role that is typically
assumed in their domains.
In perception, as we have seen, truth (as traditionally conceived, in terms of repre-
sentational accuracy or correspondence) is often considered important because per-
ceptual content, categorizations and judgments are evaluated in terms of how true
they are—e.g., how accurate the internal model is as a representation of the outside
environment. But direct realism rejects the notion of perceptual content and, through
this, it circumvents this supposedly foundational role of truth. Categorization or judg-
ments can be evaluated in terms of truth or falsehood, but they are activities people
engage in based on their perceptual experience rather than the other way around (i.e.,
rather than their perceptual states being the result of categorization or judgments).
Still, some kind of normativity is still present in the direct realist picture which can,
more broadly construed, be described and interpreted in terms of truth, especially
when it comes to our descriptions of perceptual events and perceptual knowledge. A
classic example in ecological psychology is a shark that relies on its ability to detect
shes’ self-generated electromagnetic information to approach and hunt them (see
Turvey et al., 1981). If some scientists use an electronic device to generate equivalent
electromagnetic information in its environment, the shark will likely approach and
try to hunt the electronic device. In this case, we can say that the shark was prop-
erly detecting perceptual information, even if its action of approaching and hunting
was not properly informed by the true environment. Due to the articiality of the
environment, the available information was not specic in the same way that it is in
the shark’s typical environmental niche. Similarly, in the Ames Room—a room that,
due to its shape, distorts the size relationships of the objects inside it when looked
at from a particular point of view—we can say that the perceiver is not seeing the
true environment even though her perceptual state is a function of the available per-
ceptual information. The articiality of the environment, this time combined with a
restriction of the perceiver’s movements that forces her to look at the Ames Room
from just one point of view (see Runeson, 1988), is again what introduces an element
worthy of an evaluation in terms of truth and falsehood. In both examples, the notion
of truth can be applied when evaluating the whole situation and the actions of the
perceiver in her environmental niche are taken into account. On one interpretation or
1 3
Page 17 of 24 35
European Journal for Philosophy of Science (2024) 14:35
manner of describing the situation, for instance, the electromagnetic pattern detected
by the shark can be said to constitute a false signal, one that fails as a guide to action
(namely, the action of hunting for food), yet this doesn’t entail a failure of percep-
tion (e.g., the falsehood of the shark’s perceptual content, or the inaccuracy of the
shark’s internal model): the shark’s action was appropriate given the pattern—what
was inappropriate was the articial manipulation of that energy pattern by humans,
violating the lawful relation that otherwise holds in the shark’s niche. And the same
applies to the Ames Room: given the subjective impression that common objects in
the room have abnormal sizes, the perceiving subject’s claim that one person is twice
as tall as the other one is, of course, subject to evaluation in terms of truth values; and
as for the perceptual state itself (rather than the judgment), although it could poten-
tially be interpreted as evidence of a perceptual failure (e.g., the falsehood of the
visual information), a dierent interpretation, favored in ecological psychology, is
that, on the contrary, the subjective experience is evidence that the laws of ecological
optics do hold and that only the imposition of articial constraints, not typically part
of the niche, (here, a restriction in movement, which is a major disruption of the usual
organism-environment interaction) could make it seem otherwise. Truth is not at the
forefront of perception, and therefore it does not apply to perceptual information or
perceptual states in the traditional fashion, but it does apply to dierent activities that
are constituted or at least crucially depend on perceptual knowledge.
Much the same happens, in the domain of science, where the tension between
realism(s) and anti-realism(s) typically hinges on how optimistic or pessimistic one is
about science’s prospect of achieving truth in the way theories, models and explana-
tions represent the world. Causal pattern realism avoids this tension by recognizing
that the primary epistemic goal of scientists is understanding rather than truth. Yet
this doesn’t mean that truth never matters. The emphasis on understanding rather
than truth, and on causal patterns rather than on target phenomena as a whole, does
not preclude the existence of particular contexts in scientic practice where scientists
care about truth about target phenomena. Accurately representing some biological
or physical system, for instance, or linguistically describing that system in as much
detail as possible, may well be important for some purposes in some contexts—and
in those particular contexts, failure to deliver truth would be a signicant shortcom-
ing. No doubt, success in these cases is appropriately determined in terms of the
truth value of the representations, descriptions, and so on. The observation motivat-
ing causal pattern realism is that these sorts of activities and goals are not the most
fundamental activities and goals. When it comes to explanation-seeking investigation
practices, especially in model-based research, scientists often willingly compromise
on truth and accuracy broadly construed, thus revealing that these are subsidiary to
what ultimately matters: understanding. And while representations, depictions and
declarative sentences about real-world systems and phenomena as statements of what
is known can be judged as better or worse in terms of how well they describe those
systems and phenomena, it would be wrong to mistake these facts for the means
by which knowledge was obtained. In many cases, the means to understanding was
intervention on models that embody some causal patterns of interest but which, if
evaluated in terms of overall accuracy in representing the target phenomenon as a
whole, would be properly described as more false about the target than other compet-
1 3
35 Page 18 of 24
European Journal for Philosophy of Science…
ing models. The fact that scientists routinely prefer models that are representation-
ally defective (or “false”) indicates that interpretation of epistemic success in terms
of representation with truth values (i.e., the “if” in the previous sentence), although
popular in philosophical circles, is an external imposition, one that violates what
scientists themselves see as constituting epistemic success. The mistake, that is, is
to think that the domains where truth matters more than anything else are more fun-
damental, maybe even instrumental for understanding, and accordingly to interpret
and evaluate the tools of science (e.g., models) in terms of truth value: as Potochnik
compellingly argues, in many cases in science, and perhaps more often than not,
understanding of causal patterns is the primary goal, and when truth about target
phenomena in general gets in the way, it is willingly given up on.
Direct realism and causal pattern realism thus coincide in two key conceptual
moves, rst, coinciding in the fact that they depart from the emphasis on truth that
is typical in their domains, and, second, coinciding in how they do this, namely by
redescribing the object of interest in their domain. Having spelled out the similarity in
more detail, we can now move to articulating the deeper connection we see between
the two forms of realism. Our proposal is that these two classes of conceptual moves
belong to a genus of realism—a genus we have named pragmatist realism. Pragma-
tist realism is a general realist attitude founded on the redescriptions of the general
aims and targets of knowledge and of the role of truth in them.4 When such rede-
scriptions are in place with regard to perception, a species of direct realism emerges.
When they are in place with regard to scientic models and theories, a species of
causal pattern realism is developed. But why do we choose pragmatism to label this
genus of realism? There are both historical and epistemological reasons to do so.
Historically speaking, and specially through the works of Dewey (1929), pragmatism
is associated with the scientic instrumentalism that, subsequently, was the general
theoretical framework in which some of the ideas supporting causal pattern realism
emerged. At the same time, ecological psychology is a scientic project that directly
emerges from William James’ pragmatism and radical empiricism (Costall, 2023;
Heft, 2001; Segundo-Ortin & Raja, 2024). William James was the advisor of E. B.
Holt, an American psychologist who was both a member of the self-proclaimed new
realists (Charles, 2011) and the main inuence of James Gibson’s scientic train-
ing in psychology (Gibson, 1967b). This biographical connection furnishes all the
aspects of direct realism.
The epistemological inuence of pragmatism both on direct realism and on causal
pattern realism is manifest in the conceptual moves explored in the previous two sec-
tions. Pragmatism is the source of the two re-descriptions at the basis of both kinds
of realism. At the end of the day, James takes the scope of pragmatism to be “rst,
a method; and second, a genetic theory of what is meant by truth” (1907, p. 65–66;
4 We choose this name to signal our proposal as dierent from other proposals that aim to tie pragmatism
and realism together. An example of these proposals is pragmatic realism as defended by Torretti (2000)
or Chang (2016), for instance. There are similarities between pragmatic and pragmatist realism that have to
do mostly with their treatment of truth. We will not present these similarities here. There are as well dier-
ences that drive our divergence in naming. The main dierence has to do with the restricted focus of prag-
matic realism on the notion of truth. Although, as already noted, we also have explored the notion of truth,
pragmatist realism is a wider conception of the inuence of pragmatism in the presented realist views.
1 3
Page 19 of 24 35
European Journal for Philosophy of Science (2024) 14:35
see also Sanches de Oliveira, 2022b). Regarding truth, pragmatists abandon the cor-
respondence criterion (alluded to above) and defend a notion of truth tied to practice:
“truth” is the name for what makes our practices successful—and, as causal pattern
realism emphasizes, sometimes our practices are more successful if and when we
work with what, under the narrower traditional rubric (in terms of correspondence, or
representational accuracy), would be described as falsehoods. This pragmatist notion
of truth is the one at play in direct realism and causal pattern realism. In both frame-
works, truth gets reframed in such a way as to become entangled with particular prac-
tices—either scientic or behavioral—and their success. Truth (as correspondence)
can be relevant for a concrete aim within a given scientic eld and the practices
associated to it just as it can be relevant for understanding what kind of practices
(e.g., actions) in a given articial environment, such as the Ames Room, are the right
ones to not to be deceived; yet these cases fall under the umbrella of truth in the more
general, pragmatist sense as practical success.
Eectively, the pragmatist notion of truth allows for a plurality of truths within
the same world and situation depending on the practical contexts in which they are
developed. In consonance with this line of thought, pragmatism is characterized as:
The attitude of looking away from rst things, principles, ‘categories,’ sup-
posed necessities; and of looking towards last things, fruits, consequence, facts.
(James, 1907, pp. 54–55).
The method of pragmatism is, therefore, a re-description (or perhaps a re-building, a
re-construction) of the usual methods of classical epistemology. Pragmatism brings
inquiry to the fruits and consequences of knowledge instead of its necessities and
principles. This allows for philosophical and scientic theories to be instruments (see
James, 1907, p. 53) and not reections of the one and only true world. Indeed, as dif-
ferent practices may entail a diversity of consequences and fruits, a plurality of truths
can coexist based on the dierent aims of the agents that engage in those practices—
just like dierent perceptual states may emerge in similar environmental contexts
depending on the actions of the organism and dierent scientic models of the same
target phenomena can be more epistemically successful depending on the scientic
aims of the researcher.
Pragmatism fully accommodates variability and plurality in knowledge. In this
sense, pragmatism is compatible with instrumentalist ideas regarding both scientic
modeling and cognition (e.g., Sanches et al., 2021). However, pragmatism makes this
possible without necessarily falling into ctionalism or relativism. Although some
neo-pragmatists, like Richard Rorty, openly embraced a form of relativism, prag-
matism as such can be read in a realist way. James and Dewey, for example, thought
that both the instruments used and the practices performed by agents to cope with
their world are means to access real aspects of the latter. Moreover, both James and
Dewey dismissed intellectualist theories and models that, according to them, were
not able to capture the reality of their phenomena of interest—e.g., James’ accusa-
tion of inferential theories of perception to be “pure mythology” in The Principles
of Psychology (1890). Pragmatism and realism are compatible in dierent ways (see
Torretti, 2000; Chang, 2016) and pragmatist realism is one result of this compat-
1 3
35 Page 20 of 24
European Journal for Philosophy of Science…
ibility. Pragmatist realism allows for embracing both realism and instrumentalism at
the same time while rejecting ctionalism and relativism regarding perception and/
or scientic modeling: the opposite of pragmatist realism is not instrumentalism, but
intellectualism and veritism (i.e., when narrowly construed—as it typically is—in
terms of truth as correspondence or representational accuracy).
Direct realism and causal pattern realism are species of the genus of pragmatist
realism, and they are such because their particular conceptual moves are instances of
the more general conceptual moves of pragmatism. That said, it seems to be expected
that accepting pragmatist realism entails the acceptance of both direct and causal
pattern realism. Although this is possible (and, we think, attractive), it’s not neces-
sary. One can accept direct realism without accepting causal pattern realism and vice
versa. In other words, one can think we have direct perceptual access to real aspects
of the world while rejecting such access from our scientic models and theories. And,
conversely, one could accept causal pattern realism in the domain of scientic prac-
tice, while still holding onto intellectualist and veritist assumptions about perception,
cognition, and mind. Still, the relationship between direct realism and causal pattern
realism is interesting, and, although it’s possible to accept one without the other, we
want to highlight the benets of connecting the two.
There are dierent ways of articulating this connection, with two directions of
inuence. On the one hand, we can think about ideas like the ones discussed in Sect. 2
(about perception) in light of the ideas from Sect. 3 (about science). Along these
lines, it’s possible to apply causal pattern realism to the philosophy of cognitive sci-
ence and, for instance, based on ideas concerning science and the epistemic achieve-
ments of scientists in general, we can think of psychological science as research
aiming at understanding mind and behavior through grasping causal patterns: in this
sense, then, the dierences between ecological psychology and computationalism/
representationalism can be made sense of in terms of potentially dierent causal pat-
terns and/or potentially dierent inquiry projects, with distinct aims with regard to
understanding and explanation. Some aspects of this connection have been sketched
in the recent literature (see, e.g., Sanches de Oliveira, 2023), but many important
details are still missing. Being clear on the relation between the two realisms, as we
propose in this paper, could help drive progress on this front.
On the other hand, and conversely, we want to propose that it’s also possible to
think about the ideas from Sect. 3 (about science) in light of the ideas from Sect. 2
(about perception). This would constitute an extension of the ecological view, from
the usual cases of perception-action in ordinary contexts, so as to account for the
perception-action that humans engage in when doing science. Consider how, accord-
ing to ecological psychology, in daily activities we resonate to informative patterns
emerging in a dynamically unfolding organism-environment system: central to the
ecological approach to explaining psychological phenomena, therefore, is the idea
that we are sensitive to these informative patterns and that they constrain and enable
behavior (see Raja & Anderson, 2021). The crucial point for the present discussion
is that these informative ecological patterns are also causal patterns: and just as we
resonate to causal patterns in daily activities (and as other animals also do), so do we
resonate to causal patterns when doing science. This perspective echoes the well-
known classical pragmatist idea that there is no fundamental discontinuity between
1 3
Page 21 of 24 35
European Journal for Philosophy of Science (2024) 14:35
“ordinary” practices and scientic practices. In both domains, coping successfully
is what matters: questions about truth and the truth value of propositions we might
make in the context of coping become of secondary interest because they are also
ontologically secondary (or derivative) to direct perceptual attunement.
Summing up, we hope to have opened in this paper an alternative path that rec-
onciles realism and instrumentalism in the general case. As we have pointed out,
pragmatist realism is a realist alternative opposed just to intellectualism and veritism
but not to instrumentalism. In this sense, pragmatist realism is a genus of realism that
makes it possible to accommodate both realist and instrumentalist commitments in
a coherent paradigm. The two species of realism reviewed here, direct realism and
causal pattern realism, are instances of pragmatist realism in two dierent elds—
perception and science. They illustrate the possibilities of pragmatist realism both
in terms of understanding realism in dierent domains and in terms of providing a
general framework to characterize and relate dierent species of realism.
Funding Vicente Raja was supported by a “Juan de la Cierva-Incorporación” Fellowship (Grant#
IJC2020-044829-I) funded by MCIN/AEI /10.13039/ 501,100,011,033 and by the European Union
NextGenerationEU/PRTR, and by the grant PID2021-127294NA-I00 funded by MCIN/AEI/https://doi.
org/10.13039/501100011033 and by the European Union NextGenerationEU/PRTR.
Open Access funding provided thanks to the CRUE-CSIC agreement with Springer Nature.
Declarations
Conict of interest All authors declare they have no conicts of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License,
which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long
as you give appropriate credit to the original author(s) and the source, provide a link to the Creative
Commons licence, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use
is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission
directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/
licenses/by/4.0/.
References
Andrews, M. (2021). The math is not the territory: Navigating the free energy principle. Biology & Phi-
losophy, 36, 30. https://doi.org/10.1007/s10539-021-09807-0
Byrne, A. (2005). Perception and perceptual content. In E. Sosa, & M. Steup (Eds.), Contemporary debates
in Epistemology (pp. 231–250). Blackwell.
Callender, C., & Cohen, J. (2006). There is no special Problem about scientic representation. THEORIA
Revista De Teoría Historia Y Fundamentos De La Ciencia, 21(1), 67–85.
Castellini, C., Tommasi, T., Noceti, N., Odone, F., & Caputo, B. (2011). Using object aordances to
improve object recognition. IEEE Transactions on Autonomous Mental Development, 3(3), 207–215.
Chang, H. (2016). Pragmatic realism. Revista De Humanidades De Valparaiso, 8(2), 107–122.
Chang, H. (2022). Realism for realistic people. Cambridge University Press.
Charles, E. P. (Ed.). (2011). A New look at realism. Transaction Publishers.
Chemero, A. (2009). Radical embodied Cognitive Science. MIT Press.
Clark, A. (2015). Surng uncertainty. Oxford University Press.
1 3
35 Page 22 of 24
European Journal for Philosophy of Science…
Costall, A. (2023). Contrary imaginations: Radical empiricism or pragmatism? In V. Raja, M. Segundo-
Ortin, & M. Heras-Escribano (Eds.), Placing psychology in Context: Essays in honor of Harry Heft.
Routledge.
Dame, R. F., & Prins, T. C. (1997). Bivalve carrying capacity in coastal ecosystems. Aquatic Ecology, 31,
409–421. Kluwer Academic Publishers.
Desai, M. (1984). An econometric model of the share of wages in national income: UK 1855–1965. In
Nonlinear Models of Fluctuating Growth: An International Symposium Siena, Italy, March 24–27,
1983 (pp. 253–277). Springer Berlin Heidelberg.
Dewey, J. (1929). Experience and nature. George Allen & Unwin.
Di Paolo, Ezequel, A., Buhrmann, T., & Barandiaran, X. E. (2017). Sensorimotor life: An enactive pro-
posal. Oxford University Press.
Elgin, C. Z. (2004). True enough. Philosophical Issues, 14(1), 113–131.
Elgin, C. Z. (2017). True enough. MIT Press.
Gibson, J. J. (1961). Ecological optics. In E. S. Reed, & R. Jones (Eds.), Reasons for realism (pp. 61–75).
Lawrence Erlbaum. (The re-edition was published in 1982).
Gibson, J. J. (1966). The senses considered as perceptual systems. Houghton Miin.
Gibson, J. J. (1967a). New reasons for realism. Synthese, 17, 162–172.
Gibson, J. J. (1967b). Autobiography. In E. S. Reed, & R. Jones (Eds.), Reasons for realism (pp. 7–22).
Lawrence Erlbaum. (The re-edition was published in 1982).
Gibson, J. J. (1979). The ecological approach to visual perception. Houghton Miin.
Gibson, J. J., & Gibson, E. J. (1955). Perceptual learning: Dierentiation or enrichment? Psychological
Review, 62, 32–41.
Goodwin, R. M. (1967). A Growth Cycle: Socialism, Capitalism and Economic Growth, 1967, ED. CH
Feinstein. In Essays in economic dynamics (pp. 165–170). Palgrave Macmillan UK.
Hafri, A., & Firestone, C. (2021). The perception of relations. Trends in Cognitive Sciences, 25(6), 475–
492. https://doi.org/10.1016/j.tics.2021.01.006
Heft, H. (2001). Ecological psychology in Context. Lawrence Erlbaum Associates.
Heras-Escribano, M. (2019). The philosophy of Aordances. Palgrave Macmillan.
James, W. (1890). The principles of psychology. Dover.
James, W. (1907). Pragmatism: A new name for some Old ways of thinking. Longmans, Green, and Co.
Kirchho, M. D., Kiverstein, J., & Robertson, I. (2022). The literalist fallacy and the free energy principle:
Model-building, scientic realism, and instrumentalism. The British Journal for the Philosophy of
Science. https://doi.org/10.1086/720861
Lee, D. N. (2009). General tau theory: Evolution to date. Perception, 38, 837–858.
Massimi, M. (2022). Perspectival Realism. Oxford University Press.
Matthis, J. S., Muller, K. S., Bonnen, K. L., & Hayhoe, M. M. (2022). Retinal optic ow during natural
locomotion. PLoS ONE Computational Biology, 18(2), e1009575. https://doi.org/10.1371/journal.
pcbi.1009575
McDowell, J. (1994). The content of perceptual experience. The Philosophical Quarterly, 44(175),
190–205.
Michaels, C., & Carello, C. (1981). Direct perception. Prentice-Hall.
Mohun, S., & Veneziani, R. (2016). Goodwin cycles and the US economy, 1948–2004. Mathematical
economics and the dynamics of capitalism (pp. 107–130). Routledge.
Potochnik, A. (2017). Idealization and the aims of Science. University of Chicago Press.
Potochnik, A. (2023a). Truth and reality: How to be a scientic realist without believing scientic theories
should be true. In K. Khalifa, I. Lawler, & E. Shech (Eds.), Scientic understanding and representa-
tion: Modeling in the Physical sciences (pp. 151–166). Routledge.
Potochnik, A. (2023b). Dierent ways to be a realist: A response to Pincock. In K. Khalifa, I. Lawler, &
E. Shech (Eds.), Scientic understanding and representation: Modeling in the Physical sciences (pp.
172–177). Routledge.
Potochnik, A., & Sanches de Oliveira, G. (2020). Patterns in cognitive phenomena and pluralism of
explanatory styles. Topics in Cognitive Science, 12(4), 1306–1320.
Potochnik, A., Colombo, M., & Wright, C. (2018). Recipes for science: An introduction to scientic meth-
ods and reasoning. Routledge.
Psillos, S. (1999). Scientic realism: How Science Tracks Truth. Routledge.
Raja, V. (2018). A theory of resonance: Towards an Ecological Cognitive Architecture. Minds and
Machines, 28(1), 29–51. https://doi.org/10.1007/s11023-017-9431-8
1 3
Page 23 of 24 35
European Journal for Philosophy of Science (2024) 14:35
Raja, V. (2019). From Metaphor to Theory: The role of Resonance in Perceptual Learning. Adaptive
Behavior, 27(6), 405–421. https://doi.org/10.1177/1059712319854350
Raja, V. (2021). Resonance and radical embodiment. Synthese, 199, 113–141. https://doi.org/10.1007/
s11229-020-02610-6
Raja, V., & Anderson, M. L. (2021). Behavior considered as an enabling constraint. In F. Calzavarini, & M.
Viola (Eds.), Neural mechanisms: New challenges in the philosophy of Neuroscience (pp. 209–232).
Springer.
Raja, V., & Chemero, A. (2020). In Favor of Impropriety. Commentary to Manuel Heras-Escribano’s the
philosophy of Aordances. Constructivist Foundations, 15(3), 213–216.
Rice, C. (2021). Leveraging distortions. MIT Press.
Runeson, S. (1988). The distorted room illusion, equivalent congurations, and the specicity of static
optic arrays. Journal of Experimental Psychology: Human Perception and Performance, 14(2),
295–304.
Sanches de Oliveira, G. (2022a). Radical artifactualism. European Journal for Philosophy of Science,
12(2), 36. https://doi.org/10.1007/s13194-022-00462-0
Sanches de Oliveira, G. (2022b). Pragmatism. The Palgrave Encyclopedia of the Possible. https://doi.
org/10.1007/978-3-319-98390-5241-1
Sanches de Oliveira, G. (2023). Explanatory Diversity and Embodied Cognitive Science: Reexivity moti-
vates Pluralism. In M.-O. Casper & G. F. Artese (Eds.), Methodology of Situated Cognition Research,
Studies in Brain, Mind (Vol. 23, pp. 51–76). Springer. https://doi.org/10.1007/978-3-031-39744-8_4
Sanches de Oliveira, G., Raja, V., & Chemero, A. (2021). Radical Embodied Cognitive Science and ‘Real
Cognition’. Synthese, 198(1), 115–136. https://doi.org/10.1007/s11229-019-02475-4
Sanches de Oliveira, G., van Es, T., & Hipólito, I. (2023). Scientic practice as ecological-enactive co-
construction. Synthese, 202(1), 1–33. https://doi.org/10.1007/s11229-023-04215-1
Segundo-Ortin, M., & Raja, V. (2024). Ecological psychology. Cambridge University Press. https://doi.
org/10.1017/9781009451413
Segundo-Ortin, M., Manuel, H. E., & Raja, V. (2019). Ecological psychology is radical enough: A reply
to Radical enactivists. Philosophical Psychology, 32(7), 1001–1023. https://doi.org/10.1080/09515
089.2019.1668238
Seth, A. (2021). Being you: A science of consciousness. Faber & Faber.
Shea, N. (2018). Representation in Cognitive Science. Oxford University Press.
Smith, A. D. (2002). The Problem of Perception. Harvard University Press.
Snow, J. C., & Culham, Jody, C. (2021). The treachery of images: How realism inuences brain and behav-
ior. Trend in Cognitive Sciences, 25(6), 506–519.
Torretti, R. (2000). Scientic realism’ and scientic practice. In E. Agazzi, & M. Pauri (Eds.), The reality
of the unobservable (pp. 113–122). Kluwer.
Tsao, T., & Tsao, Doris, Y. (2021). A topological solution to object segmentation and tracking. arXiv,
2107.02036. https://arxiv.org/abs/2107.02036
Turvey, M. T. (2018). Lectures on perception: An ecological perspective. Routledge.
Turvey, M. T., Shaw, R., Reed, E. S., & Mace, W. (1981). Ecological laws for perceiving and acting: A
reply to Fodor and Pylyshyn. Cognition, 10, 237–304.
van Es, T. (2020). Living models or life modelled? On the use of models in the free energy principle. Adap-
tive Behavior. https://doi.org/10.1177/1059712320918678
Warren, W. H. (1998). Visually controlled locomotion: 40 years later. Ecological Psychology, 10(3–4),
177–219. https://doi.org/10.1080/10407413.1998.9652682
Warren, W. H. (2006). The dynamics of perception and action. Psychological Review, 113 (2), 358–389.
Warren, W. H. (2021). Information Is Where You Find It: Perception as an Ecologically Well-Posed Prob-
lem. I-Perception. https://doi.org/10.1177/20416695211000366
Wilford, R. E., Raja, Vicente, H., Megan, and, & Anderson, M. L. (2022). The Treachery of Images:
Objects, Pictures, Words and the Role of Aordances in Similarity Judgements. Proceedings of the
44th Annual Conference of the Cognitive Science Society (pp. 3065–3971). Cognitive Science Soci-
ety. https://escholarship.org/uc/item/9sz5t2gn
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps
and institutional aliations.
1 3
35 Page 24 of 24
