Why do we mindread?

Tadeusz Zawidzki, GWU, Philosophy, MBEC, [email protected]
KNEW 2013, Kazimierz Dolny, Poland
What I mean by “mindreading”
What I mean by “why”
The received view and its discontents
An alternative proposal
Addendum on Sellars’s Myth of Jones
1.1 The Varieties of Mindreading
Behavioral generalization
Sensitivity to “low-level” mental states, like emotions,
perceptions, motor intentions
Sophisticated behavior reading: sensitivity to
rational patterns or intentional/teleological stance
Full-blown propositional attitude attribution (“as
 Jakub
believes/desires/hopes/thinks/wishes that he
was napping
1.2 Mindshaping
Mameli (2001); Zawidzki (2008;2013)
Mechanisms and practices that make likely interactants
more familiar and hence more easily, mutually
High fidelity imitation
 Pedagogy
 Norm institution and enforcement
 Self- and group-constituting narrative ideals
Distinctive of humans
Enhances the efficacy of relatively simple mindreading
1.3 The Intentional Stance
• Assume system is rational: will pursue goals by most
efficient means given available information
• Dennett glosses “goals” and “available information” as
“desires” and “beliefs”, but it isn’t what most mean by
these mental states:
“I have insisted that far from being most perspicuously
treated as (1) discrete, (2) semantically interpretable
states, (3) playing a causal role, the beliefs and desires
of the folk psychological craft are best viewed as
abstracta – more like centers of gravity or vectors than as
individualizable concrete states of a mechanism” (1998,
• Chess (Dennett 1978, 4-7)
• Infants do it (Gergely & Csibra’s “teleological stance”
(TICS 2003))
• Chimps do it (Wood & Hauser 2008)
• A theory of abstract patterns of observable behavior
No behavioral appearance / mental reality distinction
1.4 Full-Blown Propositional Attitude
Beliefs, desires, etc.
Focus of most philosophy and psychology of social cognition
 False belief task
 Epistemology, philosophical theories of practical rationality
To attribute these, one must:
Appreciate modes of presentation (opacity, intensionality)
 Appreciate complex, holistically-constrained relation to
behavior/observable circumstances
 Appreciate behavioral appearance / mental reality
 Think of behavior as caused by unobservable,
representational states
2.1 Answering “Why?” Questions:
Evolutionary Raisons D’Être
“Why?” means different things:
An individual’s conscious reason for engaging in some
 An individual’s unconscious reason for engaging in some
 Normative constraints on actions: reasons for which, ideally,
someone should engage in some activity
Evolutionary raison d’ être
How did some trait/behavior originate, and why has it
remained stable against mutations?
 Why, e.g., don’t Asperger’s patients take over human
3.1 The Received View
Coordination on cooperative projects was key to human evolutionary success
This requires accurate behavioral prediction and free rider prevention
Predicting who will/won’t renege
Predicting complementary courses of action among the cooperative
Individuals capable of accurately attributing propositional attitudes were and are
better at this
Other primates almost entirely hostile when interacting with unfamiliar individuals
Policing free riders requires identifying the insincere
Behavioral prediction improved via knowledge of causes of behavior (beliefs & desires)
Kovacs et al. (2010):
Humans are guided by internal states such as goals and beliefs. Without an ability to infer
others’ mental states, society would be hardly imaginable. Social interactions, from
collective hunting to playing soccer to criminal justice, critically depend on the ability to
infer others’ intentions and beliefs. Such abilities are also at the foundation of major
evolutionary conundra. For example, the human aptitude at inferring mental states might
be one of the crucial preconditions for the evolution of the cooperative social structure in
human societies. (1830)
3.2 Four Problems for the Received
We’re the only ones
Machiavellianism isn’t second nature
Attributing propositional attitudes unlikely to help
in coordination b/c it generates infinite regress
3.2.1 Holism
Belief/desire pairs lead to specific behaviors only against a background of
appropriate other propositional attitudes (Morton 1996; 2003; Bermudez
2004; 2009)
Suppose I believe there is tea in the cupboard and I desire a cup of tea; this
does not mean I will attempt to retrieve the tea from the cupboard
I might believe there is also a poisonous spider in the cupboard, or that I will be
punished if I open the cupboard, or I might desire something else more strongly than
tea at the moment, etc.
False belief task: even if Sally believes the treat is where Anne moved it while
she was away, she might not act on this belief (Apperly 2011)
Raises issues of computational tractability: too time-consuming to rule out all
possibilities in time to predict and coordinate successfully
Explains why other intelligent social mammals don’t bother attributing
propositional attitudes
Big problem for the received view: attributing propositional attitudes doesn’t
necessarily lead to more accurate and timely behavioral prediction
3.2.2 Other Intelligent Social Mammals
Humans are the only animal known to attribute full-blown
propositional attitudes (Call & Tomasello 2008)
Chimpanzees, Scrub-Jays, possibly cetaceans and other primates and
corvids are very sensitive to what others see or have seen, to behavioral
patterns (including abstract, rational ones), to emotions
But no evidence that they can attribute full-blown propositional attitudes:
unobservable, representational causes, with variable modes of
presentation and complex, holistically-constrained connections to
But many other species are intelligent and intensely social
Chimpanzee politics (de Waal)
Enhanced behavioral prediction would be a boon
If propositional attitude attribution exists in humans due to help with
behavioral prediction then why doesn’t it exist in other species?
3.2.3 We’re Not Good Machiavellians
One version of the received view: Machiavellian Intelligence Hypothesis (Byrne &
Whiten 1988)
Predicts human facility with higher-order intentional states
Arms race of deception and deception-detection
I know that she believes that I want her to think that I don’t know what she plans to do
But such reasoning does not come easily to human beings
Kinderman et al. (1998):
Subjects read stories that required the attribution of higher orders of intentionality, as well
as complex, non-psychological causal sequences, and were then tested for recall of such
attributions with forced choice questions. Only recall of attributions of the lowest order of
intentionality, e.g., attributions of belief about non-psychological states of affairs, showed
error rates comparable to recall of correspondingly complex non-psychological
They conclude that the attribution of higher orders of intentionality shows a great deal of
variation in normals, and probably places strong demands on domain-general capacities
like executive functioning, working memory, and attention
3.2.4 PA Attribution Won’t Help
Familiar infinite regress for “rational” solutions to
coordination problems (Lewis 1969; Gilbert 1996;
Bacharach 2006)
Interrupted phone call
should call back first only if she believes that I should
call back first, but she believes that I should call back
first only if I believe that I should call back first, so I
should call back first only if she believes that I believe
that I should call back first, etc.
3.2.5 Propositional Attitude Attribution
Is Not Obviously Necessary
Keep in mind all the other mindreading capacities that
we share with non-human animals:
Behavioral generalization
 Sensitivity to low-level mental states
 Intentional stance
Add to this sophisticated mindshaping (which we don’t
share with non-human animals but which enhances
capacities we do share with them)
Given this, and the 4 problems with using propositional
attitude attribution as a means of behavior prediction,
its evolutionary origin and continued stability is a
4.1 What’s the Alternative?
Propositional attitude attribution long recognized to
play justificatory as well as predictive/explanatory
Many philosophers of mind assume the former is
primary and the latter derivative (Fodor)
Maybe it’s the other way around
Full-blown propositional attitude attribution arises
only in populations that already have normative
practices relative to which behavior must sometimes
be justified by appeal to non-obvious mental states
4.2 The Evolution of Language
Agrammatical proto-language consisting of 2-word, unstructured sequences
composed of proto-names and proto-predicates (Bickerton 1990):
Origins of grammar?
“Food bring!”
Recursion, structural complexity
Functions to express prior, recursive, structurally complex thought (Pinker & Bloom
Birdsong: a model of near-recursive, structurally complex communication
that doesn’t serve such an expressive function (Fitch 2010)
Analog in human prehistory: advertising reliability & competence at coordination
on cooperative projects
Costly signaling of commitments to cooperative projects that puts status on the
4.3 Two Interpretive Strategies
Once such a complex signaling system is on the scene,
there are two potentially competing interpretive
strategies available:
Intentional stance
 Taking people at their word
Neither guarantees success, but raises the question:
What do they really think?
Introduces distinction b/w behavioral appearance and
mental reality
Being wrong about appearances is different from
appreciating reality behind appearances
4.4 PA Attribution & In/Exculpatory
Jerome Bruner (1990):
The Knobe effect (Pettit & Knobe 2009; Knobe 2003; 2006):
… when you encounter an exception to the ordinary, and ask somebody what is
happening, the person you ask will virtually always tell a story that contains
reasons (or some other specification of an intentional state) … All such stories
seem to be designed to give the exceptional behavior meaning in a manner that
implicates both an intentional state in the protagonist (a belief or desire) and
some canonical element in the culture … The function of the story is to find an
intentional state that mitigates or at least makes comprehensible a deviation from
a canonical cultural pattern. (49-50, original emphasis)
Did the CEO intentionally help/harm environment? Help: no. Harm: yes.
Indeterminacy about mental state resolved by appeal to normative status
Function in prehistory:
Apparent reneging on commitment (“The herd is north of here”) mitigated by
appeal to propositional attitude (“I thought I had been heading north; actually it
was east”)
4.5 Reason Vs. Causal History
Why did she do poorly on the test?
Fact: people much more likely to explain their own behavior in terms of reasons and others’
behavior in terms of causal history (Malle et al. 2007)
Why? 2 hypotheses:
Reason: She didn’t have time to study because she wanted to help her grandmother move instead
Causal history: Her dad’s alcoholism prevented her from developing good study habits in childhood
We know more about the mental causes of our own behavior (epistemic function)
We care more about justifying our own behavior in order to maintain status (social function)
Crucial experiments (Malle et al. 2007):
Subjects asked to explain the behavior of (1) persons with whom they were intimately acquainted, like
friends or family, (2) strangers the behavior of which they personally witnessed, and (3) strangers the
behavior of which they merely heard about
Subjects were just as likely to offer causal history explanations of the behavior of intimates, the
behavior of strangers that they had witnessed, and the behavior of strangers about which they had
Self-other asymmetry in providing reason vs. causal history explanations disappears when subjects are
motivated to portray others’ behavior in a positive light
4.6 Culturally Afforded (Self-)
Interpretive Narratives & Coordination
When we find ourselves engaged in some behavior, there are neural
mechanisms that continually and automatically generate different
construals of the imperfect behavioral data to which we have
access, searching for interpretations capable of justifying the
behavior in the eyes of our community.
Such mechanisms also automatically and continually try out different
construals of others’ observed behavior to determine status.
These processes draw on a relatively limited palette of stereotypic,
justificatory narratives afforded by the ambient culture, which
encode the propositional attitudes that are appropriate relative to
certain behavioral patterns.
The stereotypes encoded in these narratives can then serve a
regulative, mindshaping function.
E.g., the “crush narrative”
4.7 Derivative Predictive Use of PA
If most of one’s potential interactants self-regulate in complementary
directions, coordination is dramatically facilitated.
Explains the ease with which propositional attitude attribution can
be used to predict behavior, despite the holism problem:
Because most interactants are shaped to interpret behavioral evidence
as implicating the same, or complementary, culturally salient, stereotypic
narratives, and then use these narratives to regulate further thought and
behavior, the attribution of propositional attitudes expected of
characters in such narratives can support reliable behavioral prediction.
Even if quotidian propositional attitude attribution started off
performing a non-epistemic, impression management function in
prehistory, this does not rule out the possibility that it eventually came to
perform an epistemic function
5.1 Conclusion
This is offered as a speculative alternative to the
consensus view that avoids some of the problems
with it
Meant to motivate alternative research program –
no definitive proof, just a hypothesis
“Ecumenical” potential:
 Naturalizing
 Obvious Foucauldian affinity
6.1 The Myth of Jones Revisited: Our
Dennettian Ancestors
The implausible features of Sellars’s myth are relatively superficial.
These consist in:
The assumption that our ancestors were restricted to a Rylean language
capable of representing only concrete, observable properties of behavior.
The assumption that theorizing was reliable enough in human prehistory to
yield adaptive benefits.
When these assumptions are replaced with empirically defensible alternatives,
one can generate a plausible picture of the origins of our concept of mind that
maintains the spirit of Sellars’s myth.
Rather than Ryleans, our ancestors were Dennettians: they parsed bouts of
behavior into abstract categories like “goal” and “instrumentally rational
Rather than a theoretical psychologist, Jones was an innovative advocate,
hypothesizing hidden mental states in attempts to rehabilitate status
threatened by failures to abide by discursive norms.
For More Details…
Check out my new book:

similar documents