Content uploaded by Evan Westra
Author content
All content in this area was uploaded by Evan Westra on May 20, 2020
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=cphp20
Philosophical Psychology
ISSN: 0951-5089 (Print) 1465-394X (Online) Journal homepage: https://www.tandfonline.com/loi/cphp20
When is mindreading accurate? A commentary on
Shannon Spaulding’s How We Understand Others:
Philosophy and Social Cognition
Evan Westra
To cite this article: Evan Westra (2020): When is mindreading accurate? A commentary on
Shannon Spaulding’s How�We�Understand�Others:�Philosophy�and�Social�Cognition, Philosophical
Psychology
To link to this article: https://doi.org/10.1080/09515089.2020.1765326
Published online: 13 May 2020.
Submit your article to this journal
View related articles
View Crossmark data
When is mindreading accurate? A commentary on
Shannon Spaulding’sHow We Understand Others:
Philosophy and Social Cognition
Evan Westra
Department of Philosophy, University of Toronto, Toronto, Canada
ABSTRACT
In How We Understand Others: Philosophy and Social
Cognition, Shannon Spaulding develops a novel account of
mindreading with pessimistic implications for mindreading
accuracy: according to Spaulding, mistakes in mentalizing are
much more common than traditional theories of mindread-
ing commonly assume. In this commentary, I push against
Spaulding’s pessimism from two directions. First, I argue that
a number of the heuristic mindreading strategies that
Spaulding views as especially prone to error might actually
be quite reliable in practice. Second, I argue that current
methods for measuring mindreading performance are not
well-suited for the task of determining whether our mental-
state attributions are generally accurate. I conclude that any
claims about the accuracy or inaccuracy of mindreading are
currently unjustified.
ARTICLE HISTORY
Received 5 July 2019
Accepted 8 August 2019
1. Introduction
Mindreading or theory of mind is the ability to predict and explain the
behavior of other agents in terms of their mental states. A very common
view among philosophers and cognitive scientists is that mindreading is an
essential part of ordinary human social cognition. According to this idea,
our basic abilities to understand why other people act as they do, to predict
what they will do next, and to plan our social interactions all involve
reasoning about people’s beliefs, desires, intentions, and other mental states.
Without this ability, the behaviors of others would appear to us as bizarre
and unpredictable, and even basic forms of social coordination that we take
for granted would become very difficult for us. Indeed, our advanced mind-
reading capacities are thought to be a key part of the explanation for why we
human beings have greatly surpassed our great ape cousins in the domains
of culture, language, and large-scale cooperation (Baron-Cohen, 1997;
Scott-Phillips, 2014; Tomasello, 2014). Mindreading, in other words, is
CONTACT Evan Westra evan.westra@utoronto.ca Department of Philosophy, University of Toronto,
Toronto, Canada
PHILOSOPHICAL PSYCHOLOGY
https://doi.org/10.1080/09515089.2020.1765326
© 2020 Informa UK Limited, trading as Taylor & Francis Group
our primary means for gaining the knowledge we need to succeed in the
social world.
In How We Understand Others: Philosophy and Social Cognition,
Shannon Spaulding is ambivalent about this view of mindreading
(Spaulding, 2018). On the one hand, she endorses the idea that mindreading
is a pervasive and important feature of social cognition –she dubs this the
“broad scope of mindreading claim”(Spaulding, 2018,p.9)–and forcefully
rejects attempts by other philosophers to undermine it. In this sense, her
book can be read as consistent with the mainstream view in the mind-
reading literature. On the other hand, however, she argues that her fellow
mainstream mindreading theorists have too often taken it for granted that
mental-state attribution is generally accurate –a habit that she attributes to
excessive emphasis on measures like the false-belief task and the tacit
assumption that accurate belief-attribution ought to be the primary expla-
nandum of mindreading theories (Spaulding, 2018, p. 43). In contrast,
Spaulding maintains, our mental-state attributions regularly misrepresent
the real mental states of others; indeed, Spaulding thinks that accuracy is
often not even our primary goal when we reason about the minds of others.
This pessimism about the accuracy of mindreading is a dominant theme
throughout the book, as well as in Spaulding’s earlier work (Spaulding, 2016,
2017). Drawing on a wide range of empirical findings from social psychol-
ogy (which have received surprisingly little attention from mindreading
theorists until now), Spaulding paints a picture of mindreading as
a complex, messy process that takes on different forms in different social
contexts and is regularly distorted by a variety of unreliable heuristics and
biases. This culminates in an argument for a version of the Model Theory of
mindreading (Godfrey-Smith, 2005; Maibom, 2003), which takes the het-
erogeneity and context-sensitivity of mindreading as its primary explanan-
dum, rather than its putative accuracy. Thus, while Spaulding agrees with
the received view that mental-state attribution is a central and pervasive part
of everyday social cognition, she pushes back against the idea that mind-
reading is a reliable source of social knowledge.
Spaulding also argues that this pessimism about theory of mind has
consequences for debates in mainstream philosophy. For example, she
argues that our unreliable mindreading abilities should affect how we
deliberate in contexts of apparent peer disagreement. The problem of peer
disagreement refers to contexts in which two agents who are epistemic peers
find themselves in possession of the same evidence and yet nevertheless
come to different conclusions regarding a particular question; each agent is
then faced with the question of whether she should remain steadfast in her
position (Kelly, 2011) or if she should treat her epistemic peer’s position as
evidence that she should conciliate and decrease confidence in her own
position (Christensen, 2007). To the extent that the problem of peer
2E. WESTRA
disagreement has any bearing upon how we should deliberate in real-world
scenarios (and is not a mere idealization), it presupposes that we are usually
able to recognize when we are in the company of an epistemic peer.
Spaulding suggests, however, that this assumption is not tenable in light of
her arguments about the unreliability of mindreading: not only do we
regularly overestimate our own epistemic status, we also display a wide
range of biases when it comes to judging the knowledge and competence of
others. If this is right, then most perceived cases of peer disagreement are
probably not cases of peer disagreement at all because we are not normally
able to recognize whether another person is an epistemic peer. Because most
of the biases relevant to evaluations of potential epistemic peers tend to
inflate how we view our own knowledge and competence relative to others,
Spaulding thinks that our default policy in contexts of apparent peer
disagreement should be to decrease our confidence in our own judgments
(Spaulding, 2018, pp. 81–88).
Like Spaulding, I think that the question of accuracy has largely been
taken for granted in the mindreading literature, and I agree that she has
identified an extremely important issue in the study of theory of mind.
However, I am not convinced that the empirical considerations she raises
support pessimistic conclusions about the overall reliability of mindreading.
In the first part of this commentary, I will argue that several of the efficient,
heuristic mental-state attribution strategies that Spaulding views as unreli-
able might actually produce fairly accurate mental-state attributions most of
the time. I will go on to suggest that the trade-offthat we make when we
alternate between deliberate and efficient modes of mindreading would be
better described in terms of precision rather than accuracy. In the second
half of the paper, I will turn to a question that arises from these debates
about the reliability of mindreading: how exactly could we measure the
accuracy of mental-state attribution anyway? After surveying several differ-
ent methodologies, I conclude that we don’t currently possess any reliable
means for determining whether a particular act of mental-state attribution is
accurate. A consequence of this conclusion is that any empirical claim about
the accuracy of mindreading is probably unjustified. This creates problems
for Spaulding’s pessimistic take on mindreading, but also for the optimists
about mindreading whom she criticizes.
2. Mindreading for accuracy versus mindreading for efficiency
A central part of Spaulding’s critique of traditional theories of mindreading
like the theory-theory and the simulation theory is that they all tacitly
assume that the process of mental-state attribution is uniform and consis-
tent across mindreaders, targets, and social contexts. When explaining how
mindreading works in, for example, the controlled experimental conditions
PHILOSOPHICAL PSYCHOLOGY 3
of the false belief task, this is excusable, but in real-world scenarios,
Spaulding argues, the mindreading process is influenced by a myriad of
factors that affect what information it takes in and how that information is
processed. As a result, mental-state attribution can end up functioning very
differently in different kinds of social contexts, and different individuals in
the same context may well arrive at divergent mental-state attributions. This
variability is a key component of Spaulding’s case for pessimism about the
accuracy of mindreading: different people in the same situations with the
same evidence can nevertheless arrive at vastly divergent psychological
interpretations of a particular individual’s actions (Spaulding, 2017,
2018, p. 36).
One factor that can influence the cognitive strategies and resources that
we employ while mindreading is our social goals. Sometimes, we are very
motivated to make correct mental-state attributions and behavioral pre-
dictions about others (say, when interviewing a potential nanny for one’s
child). In these cases, where mindreading “aims at accuracy,”Spaulding
suggests that we tend to engage in conscious, deliberative forms of reason-
ing that are cognitively taxing and liable to break down under a cognitive
load (Spaulding, 2018, p. 45). At other times, a person’sactualmental
states are not particularly relevant to our plans (e.g., we do not need to
know what a cashier at a grocery store is really thinking in order to buy
eggs and milk). In these cases, we do not engage in effortful forms of
reasoning about mental states. Instead, we employ fast and efficient mind-
reading strategies that require less cognitive effort but are more prone to
error. Thus, whether or not we care about getting our mental-state attri-
butions right can dramatically affect how mindreading functions in a given
situation.
The particular type of mindreading strategy we employ when we are
motivated by efficiency may depend upon how we have socially categorized
the mindreading target (Spaulding, 2018, pp. 45–46). If we see the target as
similar to us in some respect (e.g., as part of our in-group), we are more
likely to employ a mental-state attribution heuristic called egocentric projec-
tion. Egocentric projection occurs when a mindreader assumes that her
target shares her mental states or otherwise uses her own mental states as
an initial anchor from which she can subsequently adjust (Ames, 2004a;
Epley et al., 2004). The downsides of this heuristic are revealed whenever the
target is not actually that similar to the mindreader. In these cases, the
mindreader may infer that certain propositions are common knowledge
when they are actually not, or she may mistakenly come to believe that
certain goals are shared when they are not. Even when a mindreader does
recognize that a target is not entirely similar to herself, using her own
mental states as an anchor may nevertheless lead her to underestimate the
gap between her target’s beliefs and her own.
4E. WESTRA
On the other hand, if a mindreader sees her target as different from
herself, she may instead draw on stereotypes about the target’s perceived
social category (Ames, 2004b). As Spaulding notes, we tend to automatically
categorize people upon first encountering them. These categorizations acti-
vate stereotypes associated with a particular group, which often include
beliefs about mental states and traits that members of that group are likely
to share. For example, categorizing a person as a ‘mother’might lead to the
expectation that she cares more about family than work, while categorizing
someone as a ‘millennial’might lead to the expectation that she is entitled,
self-absorbed, and overly fond of avocado toast. Because these stereotypical
beliefs are stored in an efficient format, they become readily accessible
whenever we are faced with the task of predicting and interpreting
a target’s behavior. This can lead to biased interpretations and predictions
of a target’s actions, such that the same behavior might be interpreted one
way when it comes from a mother, another way when it comes from
a millennial, and so on.
Spaulding’s insight that a mindreader’s goals and social categorizations
can affect how mentalizing functions is a good one. It is not clear, however,
that the modes of mindreading that aim for efficiency rather than accuracy
are, in fact, unreliable. Consider first egocentric projection: as noted, this
strategy will fail when it turns out that we have overestimated the similarity
between ourselves and our target. However, when it comes to the vast
majority of our everyday beliefs, we are often quite similar to the people
around us (Heal, 1996). Consider, for instance, the belief that the Allies won
World War II, the belief that dark clouds mean rain, or the belief that today
is Tuesday. Because these beliefs are so mundane, I do not need to engage in
deliberative reasoning in order to attribute them to the people around me,
nor do I need any particular reasons for these attributions beyond my
evidence for holding those beliefs in the first place. In projecting these
beliefs of mine onto others, there is vanishingly little chance of error. The
chance of error further decreases when one takes into account the fact that
people naturally sort themselves into groups of like-minded individuals, for
instance, when we go to philosophy conferences, spend time with our
friends, or cluster together online in political echo chambers. In these
familiar social contexts, we can reliably expect a wide range of facts to be
common knowledge and tailor our mental-state inferences accordingly. For
these reasons, the default assumption that our beliefs tend to be shared
(which is effectively a true-belief default) is built into a variety of different
accounts of mindreading not as a bug, but as a feature (Heal, 1996; Leslie
et al., 2004). Thus, while egocentric projection may occasionally lead to
striking errors (Epley et al., 2004), there is good reason to think of it as an
ecologically rational heuristic rather than as a bias (Goldstein & Gigerenzer,
2005).
PHILOSOPHICAL PSYCHOLOGY 5
It is also not obvious that stereotype-based mindreading is particularly
unreliable. Certainly, one can find many examples of inaccurate stereotypes
that lead to mindreading errors and perpetuate harmful social inequalities
and structural injustices. To cite just one example that Spaulding provides,
many people falsely believe that people with disabilities tend to have a lower
than average quality of life, which could, in turn, lead to a number of
misleading mental-state inferences (Spaulding, 2018, p. 47). However,
stereotypes are not false by definition. As a cognitive structure, stereotypes
are just an efficient way of storing and retrieving semantic information
associated with different social categories (Amodio, 2014). The truth or
falsity of a given stereotype thus depends upon the content that it encodes,
not on the mere fact that it is a stereotype.
1
For example, consider the stereotype that most American academics are
politically liberal, which happens to be true (Duarte et al., 2014). If one were
to rely upon this stereotype when anticipating an American academic’s
beliefs about a divisive political issue such as gun control or abortion, one
would probably arrive at the correct answer most of the time. This example
is not unique: there is a large, though controversial (see, e.g., Bian &
Cimpian, 2017) empirical literature showing that many of our demographic
stereotypes are surprisingly accurate when compared to real-world statistics
(Jussim, 2012; Jussim et al., 2015). Relying upon any of these stereotypes
during mental-state attribution would therefore seem like a fairly safe bet,
provided one does not have any additional individuating information about
the mindreading target (Crawford et al., 2011).
This is not to deny that the stereotype-based errors that Spaulding cites
occur, nor that they are not morally worrisome: they are certainly both.
Even when they are true, relying upon stereotypes can be morally proble-
matic, both because doing so perpetuates the underlying social injustices
that make them true, and because treating a person as a mere instance of
a kind seems to reflect a failure to respect that person as a unique individual.
However, when we are trying to understand the relationship between
stereotyping and mindreading, and trying to understand the role of stereo-
types in social cognition more generally, we should nevertheless be careful
not to treat cases of error as representative. The ubiquity of stereotyping in
social cognition does not imply the ubiquity of erroneous stereotyping.
Focusing on these cases obscures the fact that stereotype-based mindreading
often works quite well.
The fact that mindreading processes that “aim for efficiency”may actually
be quite reliable problematizes Spaulding’s analysis of modes of mindread-
ing that “aim for accuracy.”It is certainly true that we are sometimes
motivated to engage in conscious, deliberative forms of mindreading that
are more effortful than stereotyping and projection, but it is not clear that
this effort buys increased accuracy. To better characterize the function of
6E. WESTRA
this mode of mindreading, we must distinguish between accuracy and
precision. To illustrate, consider the contrast between the following two
predictions:
(1) It will rain in July.
(2) It will rain in the morning on July 3.
If it rains in the morning on July 3, then it will be the case that both
predictions were accurate. However, the two predictions nevertheless
differ in terms of their precision. The first prediction is relatively easy to
make, but it is also quite imprecise: it doesn’t tell us when in July it will
rain. This limits its utility because it does not help us make specificplans,
such as deciding on which day we should have a barbecue or an outdoor
wedding. In contrast, the second prediction is very precise and would
support highly specific plans. Making it with any confidence would, how-
ever, require more cognitive effort, and it would have a higher chance of
being wrong.
I suspect that the difference between efficient and deliberative forms of
mindreading involve a similar trade-offbetween cognitive effort and preci-
sion. While heuristics like egocentric projection and stereotyping may lead
to reliably accurate mental-state attributions, they are also relatively impre-
cise. This might be fine as long as our plans do not depend upon having
a very precise representation of a person’s mental states –for instance, when
we are interacting with a cashier at the checkout counter or chatting with
a stranger while waiting for a bus. Relying upon egocentric projection or
stereotyping in these contexts may lead us to construct informationally
sparse models of other agents that fail to capture many details about their
actual attitudes. In many cases, however, a person’s specific mental states do
matter to us, and it is worthwhile for us to engage in explicit perspective-
taking in order to understand what they are thinking –say, during a serious
conversation with a romantic partner or when a therapist attempts to
understand her patient’ssuffering. By consciously reasoning about
a person’s mental states and imagining what their experiences are like, we
buy ourselves an added degree of precision that can be crucial when
negotiating significant interpersonal relationships or coordinating on com-
plex plans (Westra, 2018,2019b).
In short, while our goals and social categorizations might influence the
kind of mindreading strategy we adopt, this does not give us reason to be
worried about the overall accuracy of mindreading. Efficient mindreading
strategies may trade offon precision, but they can nevertheless function as
ecologically rational and reliable heuristics. They need not be viewed as
“biases”in an epistemically pejorative sense, even though they do result in
morally problematic judgments some of the time.
PHILOSOPHICAL PSYCHOLOGY 7
3. How do we measure accurate mental-state attribution?
Spaulding’s case for inaccuracy in mindreading is not limited to stereotyp-
ing and egocentric projection, however. In her book, she cites a long list of
domain-general biases discovered by social psychologists, such as naïve
realism (Pronin et al., 2002), self-serving and group-serving attribution
biases (Miller & Ross, 1975; Pettigrew, 1979), and the Dunning-Krueger
effect (Kruger & Dunning, 1999), and she describes how these might
negatively impact the accuracy of mental-state attribution –for example,
by distorting our appraisals of others’knowledge and competence relative to
our own (Spaulding, 2018, pp. 81–85). These, she might argue, are genuine
biases, and they seem much less amenable to an explanation in terms of
ecological rationality. Should not the existence of such a wide range of biases
cause us to be pessimistic about the accuracy of mindreading and other
social judgments?
2
Such pessimism would be too hasty, however. The mere fact that these
biases are likely to lead to mindreading errors does not tell us much about
the overall frequency of those errors relative to accurate mental-state attri-
butions. Even the fact that a large number of these biases exists does not tell
us about their overall effects on mindreading. Consider an analogous argu-
ment, in which I try to argue that human vision is unreliable by listing off
the dozens of optical illusions to which we are prone. This kind of argument
would be unconvincing because we know that these kinds of illusions
comprise only a minority of our perceptual experiences and that our visual
capacities are generally reliable despite them. Likewise, the fact that we are
prone to systematic patterns of error in certain contexts is consistent with
the claim that overall, mindreading is reliable and trustworthy (Westra,
2019a). Thus, in order to make this claim –that mindreading is, overall,
accurate –it does not suffice to provide a list of all the biases that might
cause mindreading errors. What is really needed is a way of measuring the
accuracy of mental-state attribution in its own right.
Problematically, many of the most commonly used mindreading para-
digms are not actually concerned with measuring accuracy at all. For
example, developmental theory-of-mind paradigms such as the false-
belief task (Wimmer & Perner, 1983) and the theory-of-mind scale
(Wellman & Liu, 2004) only purport to measure whether a participant
possesses certain mental-state concepts. In tasks like these, young children
are asked to make basic action predictions based on what fictional char-
acterswant,think,orknowinaparticularscenario;succeedinginthese
tasks requires children to understand that other agents can have mental
states that differ from their own (e.g., that a character with a false belief
might search for an object in a location the child knows to be incorrect).
In order to isolate conceptual competence from extraneous performance
8E. WESTRA
factors, these paradigms are intentionally designed to be as simple as
possible and do not reflect the complexities of real-world social
interactions.
3
Of course, possessing mental-state concepts is a necessary
condition for accurate mental-state attribution: one cannot correctly
attribute beliefs unless one knows what beliefs are. However, beyond
establishing a basic prerequisite for accuracy, these tasks do not tell us
anything about the broader reliability of mindreading.
Tasks that are used to measure more advanced mindreading abilities in
adults are likewise unsuited for measuring accuracy. The reading-the-mind-
in-the-eyes test (RMET) (Baron-Cohen et al., 2001), for example, measures
how well people are able to apply various emotion words to specific facial
expressions. Participants in this task view a series of images depicting the
eyes and part of the face of a model and must choose which emotion word
among four options best describes the model’s current psychological state.
To succeed in the task, they must be able to make fine-grained discrimina-
tions between different facial expressions and also express those discrimina-
tions using sophisticated psychological vocabulary.
4
Again, this ability
might be relevant to the accuracy of mindreading, since it reflects partici-
pants’ability to individuate different types of psychological states according
to their observable correlates. However, being able to classify a face accord-
ing to the type of emotion it displays sets the bar for accuracy fairly low:
recognizing that a person appears to be pensive is not nearly as hard as
judging what they are being pensive about –that is, the particular content of
the emotional state in question. Even in terms of emotional categorization,
the decontextualized nature of the RMET leaves much to be desired, since
facts about the context in which a face occurs (e.g., bodily posture) can
radically change whether it is interpreted as expressing positive or negative
emotions (Aviezer et al., 2017,2012). It is thus not clear that performance on
the RMET says much at all about the accuracy of mental-state attribution.
In contrast, recursive mindreading tasks (e.g., Stiller & Dunbar, 2007)do
involve the attributions of particular contents, but not as the result of any
sort of inference. Rather, these tasks measure participants’ability to keep
track of mental states of characters from a story when this information is
presented in a complex, multiply embedded format (e.g., “Bill knows that
John thinks that Sally doesn’t know that Jenny thinks she’s nice.”). Here, the
correct mental-state information is simply given to participants in
a narrative format; the only way to be inaccurate in this task is to forget
the correct answer (hence, why Stiller & Dunbar, 2007 also found
a significant correlation between recursive mentalizing and short-term
memory). Of course, maintaining relevant information in one’s memory is
sure to be relevant to any kind of complex problem-solving, including
problems that involve mentalizing. Thus, this task also measures something
PHILOSOPHICAL PSYCHOLOGY 9
that is relevant to accurate mindreading. It is, however, certainly not a test of
mindreading accuracy as such.
The closest we have come to finding a way to measure the accuracy of
mental-state attribution comes from the literature on “empathic accuracy”
(Hodges et al., 2014; Ickes et al., 1990). In these studies, experimenters
videotape a target individual as she is being interviewed or while she is
having a conversation. Afterward, the experimenters show the tape to the
target and ask her to indicate what she was thinking at different points in
time. The same video is then shown to another participant. At different
points during the video, this participant is asked to guess what the target is
thinking. Afterward, independent coders rate how similar the judge’s
guesses are to the target’s self-reports.
5
This use of self-reported mental states as an accuracy criterion is proble-
matic for several reasons. First, it assumes that we have reliable introspective
access to our own thoughts. The problem with this is that there is a large
body of evidence from both neurotypical and clinical cases showing that we
are prone to confabulate about both our prior decision-making and our
occurrent conscious experiences (Johansson et al., 2005; Nisbett & Wilson,
1977; Roser & Gazzaniga, 2004; Schwitzgebel, 2002). This research suggests
that when targets review their recorded interviews and indicate what they
were thinking at the time, they may simply be generating plausible mental-
state attributions about themselves after the fact. This ties into a further
problem: according to theories of introspection like Peter Carruthers’inter-
pretive sensory access account, introspection is just a form of mindreading
that is directed at the self (Carruthers, 2009,2011). If this theory or some-
thing like it is correct, then the criterion which empathic accuracy research-
ers have for accurate third-party mindreading presupposes the accuracy of
adifferent kind of mindreading. Finally, the mental-state attributions that
these studies evaluate are explicit, verbal, and conscious. Yet, most mind-
reading theories hold that this kind of reasoning about other minds con-
stitutes just a small part of our overall theory of mind and that most
mentalizing is implicit, unconscious, and non-verbal. If this is right, then
it may be that we make many accurate mental-state attributions that we are
unable to put into words.
Despite these objections to the empathic accuracy approach, it is admit-
tedly difficult to come up with a criterion for accurate mental-state attribu-
tion that does not rely upon introspection. One possibility might be to
employ advanced neuroimaging methods such as multivariate pattern ana-
lysis to decode the content of a person’s occurrent mental states and then
use this as an accuracy criterion for third-party mental-state attribution (c.f.,
Ritchie et al., 2019 for some of the limitations of this technique). Barring the
significant technological and methodological advances that this would
require, introspective accuracy criteria are probably the best that we can
10 E. WESTRA
expect to achieve. If the above-mentioned concerns about introspection are
warranted, then this bodes ill for the measurement of accuracy in
mindreading.
This is shocking, considering how often mindreading theorists appeal to
the adaptive significance of mindreading and its importance in social learn-
ing. Contrary to Spaulding’s pessimistic view, however, the problem with
this assumption is not that we are less accurate than these theorists would
have us believe. It is, rather, that we are not in an epistemic position to know
whether mindreading is accurate one way or the other. In other words,
claims about the accuracy and inaccuracy of mindreading are both equally
unjustified. If this is right, then mindreading theorists cannot claim to offer
advice to philosophers who are worried about peer disagreement or about
any other philosophical issue that presupposes the accuracy of our mental-
state attributions. It could be that Spaulding is, in fact, correct that we
should not trust our judgments about who our epistemic peers are.
Alternatively, it could be that strategies like egocentric projection and
stereotyping actually make mindreading more reliable, as I argued above.
Without any way to empirically test such claims, however, neither position
enjoys any direct evidential support.
4. Conclusion
In this commentary, I have focused on one theme that emerges in Spaulding
(2018): the accuracy of mindreading. I have offered some pushback against
Spaulding’s arguments about the inaccuracy of mindreading, and then
I took a step back to point out what I think is a bigger problem regarding
accuracy. In spite of these points of disagreement, I am largely sympathetic
to Spaulding’s overall project. Theories of mindreading ought to take into
account the role that different motivational and social factors play in the way
that mindreading functions, and Spaulding’s model theory provides
a promising framework for talking about these effects. Moreover, by taking
up the issue of accuracy, Spaulding has advanced the conversation about
mindreading beyond questions about its internal and conceptual under-
pinnings to questions about how our mental-state attributions actually
reflect the way the world is. With this commentary, I hope to have built
upon Spaulding’s advances by highlighting the complexity of this issue, and
I hope to have shown that it is worthy of further philosophical attention.
Notes
1. Spaulding is sensitive to this point. In particular, she is careful to distinguish between
stereotypes that encode false statistical generalizations and stereotypes that encode
generics (Leslie, 2014). For example, “most mosquitos carry West Nile”expresses
PHILOSOPHICAL PSYCHOLOGY 11
a false statistical generalization, but the generic “mosquitos carry West Nile”expresses
a true proposition. Spaulding allows that the stereotypes that influence mindreading
may be either statistical or generic in their form, and that generic stereotypes some-
times encode true propositions.
2. Spaulding (2018) also argues that sometimes our goals in mindreading are not to
attribute accurate mental states at all but, rather, to shape or regulate the mental
states of our mindreading targets (McGeer, 2007; Zawidzki, 2013). For example, one
might try to encourage a person to sample a new kind of cuisine by convincing
them that they like to try new things. Ostensibly, this mental-state attribution need
not be true in order for it to achieve its goal. However, this goal is entirely
consistent with –and indeed might actually depend upon –accurate mindreading.
In order for subtle acts of regulative mental-state attribution like this to be effective,
however, they will most likely require other accurate mental-state attributions. For
example, in order for the aforementioned strategy to work, I must first be able to
predict that the false mental-state attribution would have a particular effect on the
target’s actual mental states (say, by activating the person’s desire to like up other
people’s expectations of her). If this part of my prediction is wrong, then the whole
regulative strategy would fail. Thus, the fact that we sometimes mindread for
regulative purposes does not undercut the basic claim that mindreading typically
aims at accuracy.
3. Indeed, a common critique of these kinds of tasks is that they involve extraneous
performance demands that obscure conceptual competence (Helming et al., 2016;
Leslie et al., 2005; Westra & Carruthers, 2017).
4. It is also worth noting that performance on the reading-the-mind-in-the-eyes task is
strongly affected by participant race, ethnicity, and education (Dodell-Feder et al.,
2020), which raises further questions about its reliability as a tool to measure mind-
reading competence.
5. Using this method, Lewis and colleagues have found that having stronger stereotypes
can, at least in certain circumstances, lead to more accurate mindreading (Lewis et al.,
2012)–a point that is consistent with the argument about stereotyping from the
previous section. However, as I go on to argue, we should be very cautious about how
much authority we give to this finding.
Disclosure statement
No potential conflict of interest was reported by the author.
Funding
This work was supported by the Social Sciences and Humanities Research Council of
Canada [PDF 756-2018-0012].
References
Ames, D. R. (2004a). Inside the mind reader’s tool kit: Projection and stereotyping in mental
state inference. Journal of Personality and Social Psychology,87(3), 340–353. https://doi.
org/10.1037/0022-3514.87.3.340
12 E. WESTRA
Ames, D. R. (2004b). Strategies for social inference: A similarity contingency model of
projection and stereotyping in attribute prevalence estimates. Journal of Personality and
Social Psychology,87(5), 573–585. https://doi.org/10.1037/0022-3514.87.5.573
Amodio, D. M. (2014). The neuroscience of prejudice and stereotyping. Nature Reviews:
Neuroscience,15(10), 670–682. https://doi.org/10.1038/nrn3800
Aviezer, H., Ensenberg, N., & Hassin, R. R. (2017). The inherently contextualized nature of
facial emotion perception. Current Opinion in Psychology,17,47–54. https://doi.org/10.
1016/j.copsyc.2017.06.006
Aviezer, H., Trope, Y., & Todorov, A. (2012). Body cues, not facial expressions, discriminate
between intense positive and negative emotions. Science,338(6111), 1225–1229. https://
doi.org/10.1126/science.1224313
Baron-Cohen, S. (1997). Mindblindness: An essay on autism and theory of mind. MIT Press.
https://books.google.com/books?hl=en&id=MDbcNu9zYZAC&pgis=1
Baron-Cohen, S., Wheelwright, S., Hill, J., Raste, Y., & Plumb, I. (2001). The ‘Reading the
mind in the eyes ’test revised version: A study with normal adults, and adults with
asperger syndrome or high-functioning autism. Journal of Child Psychology and
Psychiatry,42 (2), 241–251. https://docs.autismresearchcentre.com/papers/2001_
BCetal_adulteyes.pdf
Bian, L., & Cimpian, A. (2017). Are stereotypes accurate? A perspective from the cognitive
science of concepts. Behavioral and Brain Sciences,40,e3.https://doi.org/10.1017/
s0140525x15002307
Carruthers, P. (2009). How we know our own minds: The relationship between mind-
reading and metacognition. Behavioral and Brain Sciences,32(2), 121–138. https://doi.
org/10.1017/S0140525X09000545
Carruthers, P. (2011). The opacity of mind: An integrative theory of self-knowledge. Oxford
University Press.
Christensen, D. (2007). Epistemology of disagreement: The good news. Philosophical
Review,116(2), 187–217. https://doi.org/10.1215/00318108-2006-035
Crawford, J. T., Jussim, L., Madon, S., Cain, T. R., & Stevens, S. T. (2011). The use of
stereotypes and individuating information in political person perception. Personality &
Social Psychology Bulletin,37(4), 529–542. https://doi.org/10.1177/0146167211399473
Dodell-Feder, D., Ressler, K. J., & Germine, L. T. (2020). Social cognition or social class and
culture? on the interpretation of differences in social cognitive performance. Psychological
Medicine.https://doi.org/10.1017/S003329171800404X
Duarte, J. L., Crawford, J. T., Stern, C., Haidt, J., Jussim, L., & Tetlock, P. E. (2014). Political
diversity will improve social psychological science. Behavioral and Brain Sciences,38,
1–54. https://doi.org/10.1017/S0140525X14000430
Epley, N., Keysar, B., Van Boven, L., & Gilovich, T. (2004). Perspective taking as egocentric
anchoring and adjustment. Journal of Personality and Social Psychology,87(3), 327.
https://doi.org/10.1037/0022-3514.87.3.327
Godfrey-Smith, P. (2005). Folk psychology as a model. Philosophers’Imprint,5(6), 1–16.
https://doi.org/10.3366/jqs.2011.0012
Goldstein, D. G., & Gigerenzer, G. (2005). Models of ecological rationality: The recognition
heuristic. Psychological Review,109(1), 75–90. https://doi.org/10.1037/0033-295x.109.1.75
Heal, J. (1996). Simulation, theory, and content. In P. Carruthers & P. K. Smith (Eds.),
Theories of theories of mind (pp. 75–89). Cambridge University Press.
Helming, K. A., Strickland, B., & Jacob, P. (2016). Solving the puzzle about early
belief-ascription. Mind & Language,31(4), 438–469. https://doi.org/10.1111/mila.12114
PHILOSOPHICAL PSYCHOLOGY 13
Hodges, S. D., Lewis, K. L., & Ickes, W. (2014). The matter of other minds: Empathic
accuracy and the factors that influence it. APA Handbook of Personality and Social
Psychology, Volume 3: Interpersonal Relations. https://doi.org/10.1037/14344-012
Ickes, W., Stinson, L., Bissonnette, V., & Garcia, S. (1990). Naturalistic social cognition:
Empathic accuracy in mixed-sex dyads. Journal of Personality and Social Psychology,59
(4), 730–742. https://pdfs.semanticscholar.org/fd94/e2ae60224a
be60759a790083a922fc58b442.pdf
Johansson, P., Hall, L., Sikström, S., & Olsson, A. (2005). Failure to detect mismatches
between intention and outcome in a simple decision task. Science,310(5745), 116–119.
https://doi.org/10.1126/science.1111709
Jussim, L. (2012). Social perception and social reality: why accuracy dominates bias and self-
fulfilling prophecy. Oxford University Press. doi:10.1093/acprof:oso/
9780195366600.001.0001
Jussim, L., Crawford, J. T., & Rubinstein, R. S. (2015). Stereotype (In)accuracy in perceptions
of groups and individuals. Current Directions in Psychological Science,24(6), 490–497.
https://doi.org/10.1177/0963721415605257
Kelly, T. (2011). Peer disagreement and higher-order evidence. In A. I. Goldman &
D. Whitcomb (Eds.), Social epistemology: essential readings (pp. 183–217). Oxford
University Press.
Kruger, J., & Dunning, D. (1999). Unskilled and unaware of it: How difficulties in recogniz-
ing one’s own incompetence lead to inflated self-assessments. Journal of Personality and
Social Psychology,77(6), 1121. https://doi.org/10.1037/0022-3514.77.6.1121
Leslie, A. M., Friedman, O., & German, T. P. (2004). Core mechanisms in “theory of mind”.
Trends in Cognitive Sciences,8(12), 528–533. https://doi.org/10.1016/j.tics.2004.10.001
Leslie, A. M., German, T. P., & Polizzi, P. (2005). Belief-desire reasoning as a process of
selection. Cognitive Psychology,50(1), 45–85. https://doi.org/10.1016/j.cogpsych.2004.06.
002
Leslie, S.-J. (2014). Carving up the social world with generics. In J. Knobe, T. Lombrozo, &
S. Nichols (Eds.), Oxford studies in experimental philosophy: Volume one (pp. 208–232).
Oxford University Press.
Lewis, K. L., Hodges, S. D., Laurent, S. M., Srivastava, S., & Biancarosa, G. (2012). Reading
between the minds: The use of stereotypes in empathic accuracy. Psychological Science,23
(9), 1040–1046. https://doi.org/10.1177/0956797612439719
Maibom, H. (2003). The mindreader and the scientist. Mind & Language,18(3), 296–315.
https://doi.org/10.1111/1468-0017.00229
McGeer, V. (2007). The regulative dimension of folk psychology. In D. D. Hutto &
M. Ratcliffe (Eds.), Folk psychology re-assessed (pp. 137–156). Springer.
Miller, D. T., & Ross, M. (1975). Self-serving biases in the attribution of causality: Fact or
fiction? Psychological Bulletin,82(2), 213. https://doi.org/10.1037/h0076486
Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know: Verbal reports on
mental processes. Psychological Review,84(3), 231–259. https://doi.org/10.1037/0033-
295X.84.3.231
Pettigrew, T. F. (1979). The ultimate attribution error: Extending Allport’s cognitive analysis
of prejudice. Personality & Social Psychology Bulletin,5(4), 461–476. https://doi.org/10.
1177/014616727900500407
Pronin, E., Lin, D. Y., & Ross, L. (2002). The bias blind spot: Perceptions of bias in self versus
others. Personality & Social Psychology Bulletin,28(3), 369–381. https://doi.org/10.1177/
0146167202286008
14 E. WESTRA
Ritchie, J. B., Kaplan, D. M., & Klein, C. (2019). Decoding the brain: Neural representation
and the limits of multivariate pattern analysis in cognitive neuroscience. The British
Journal for the Philosophy of Science,00,1–27. https://doi.org/10.1093/bjps/axx023
Roser, M., & Gazzaniga, M. S. (2004). Automatic brains, interpretive minds. Current
Directions in Psychological Science,13(2), 56–59. https://doi.org/10.1111/j.0963-7214.
2004.00274.x
Schwitzgebel, E. (2002). How well do we know our own conscious experience? The case of
visual imagery. Journal of Consciousness Studies,9(5–6), 35–53. http://faculty.ucr.edu/~
eschwitz
Scott-Phillips, T. (2014). Speaking our minds: Why human communication is different, and
how language evolved to make it special (Vol. 3). Palgrave Macmillan. https://books.
google.com/books?hl=en&id=B0yjBQAAQBAJ&pgis=1
Spaulding, S. (2016). Mind misreading. Philosophical Issues, 26.
Spaulding, S. (2017). Do you see what I see? How social differences influence mindreading.
Synthese,1–22. https://doi.org/10.1007/s11229-017-1404-1
Spaulding, S. (2018). How we understand others: Philosophy and social cognition. Routledge.
https://doi.10.4324/9781315396064
Stiller, J., & Dunbar, R. I. M. (2007). Perspective-taking and memory capacity predict social
network size. Social Networks,29(1), 93–104. https://doi.org/10.1016/j.socnet.2006.04.001
Tomasello, M. (2014). A natural history of human thinking. Harvard University Press.
https://books.google.com/books?hl=en&id=ksYXAgAAQBAJ&pgis=1
Wellman, H. M., & Liu, D. (2004). Scaling of theory-of-mind tasks. Child Development,75
(2), 523–541. https://doi.org/10.1111/j.1467-8624.2004.00691.x
Westra, E. (2018). Character and theory of mind: An integrative approach. Philosophical
Studies,175(5), 1217–1241. https://doi.org/10.1007/s11098-017-0908-3
Westra, E. (2019a). Getting to know you: Accuracy and error in judgments of character.
Mind & Language.https://doi.org/10.1111/mila.12258
Westra, E. (2019b). Stereotypes, theory of mind, and the action–prediction hierarchy.
Synthese,196(7), 1–26. https://doi.org/10.1007/s11229-017-1575-9
Westra, E., & Carruthers, P. (2017). Pragmatic development explains the theory-of-mind
scale. Cognition,158, 165–176. https://doi.org/10.1016/j.cognition.2016.10.021
Wimmer, H., & Perner, J. (1983). Beliefs about beliefs: Representation and constraining
function of wrong beliefs in young children’s understanding of deception. Cognition,13
(1), 103–128. https://doi.org/10.1016/0010-0277(83)90004-5
Zawidzki, T. W. (2013). Mindshaping: A new framework for understanding human social
cognition. MIT Press.
PHILOSOPHICAL PSYCHOLOGY 15
