Topic 1: Rational Choice Theory
Rational Choice Theory models social action as the outcome of purposive,
instrumentally rational agents who hold preferences over outcomes, form
beliefs about how alternative actions map onto those outcomes, and who
operate under constraints (resources, institutions, information etc.). Given
these constraints, preferences, and beliefs about the world, actors select
the action they expect will maximizes their utility.
Preferences
‘Rational’ desires and preferences
The word ‘rationality’ invokes the idea of reason, suggesting agents who
follow the prescriptions of instrumental reason: forming justified beliefs,
endorsing appropriate ends, and choosing effective means to reach them.
However, acting in conformity with reason and acting rationally are two
different things. Conformity with reason is judged from an external
perspective, whereas rationality is judged from the agent's internal
standpoint. An action is rational if it follows coherently from the agent’s
own beliefs and preferences, regardless of whether those preferences are
themselves reasonable or well-justified. Theories of rational action
characteristically assume the agent’s desires and preferences as a fixed
given, and only evaluate how agents act given these preferences.
As a result, preferences that appear unreasonable are not necessarily
irrational. Consider a preference for consuming goods on Wednesdays as
opposed to Thursdays, solely because one prefers that day of the week.
Such a preference violates reason but does nothing to violate rational
choice theory, which places no constraints on the content of preferences.
Only incoherent desires are excluded, such as the wish that everyone earn
more than average.
Desires themselves include both substantive preferences, such as
preferring apples to oranges, and formal preferences, such as one’s
attitude towards risk, uncertainty, and the future. Consider a choice
between a guaranteed £100 or a lottery ticket offering a 0.00001%
chance of winning £1,000,000. Taking the lottery ticket may seem
irrational to most but, in expected-value terms, the two options are
equivalent. A risk-averse agent may rationally prefer the guaranteed
£100, but a risk-seeking agent may rationally prefer the lottery. There is
no uniquely “rational” option independent of the agent’s formal
preferences, because rationality concerns coherence between
preferences, beliefs, and choice - not conformity to an objectively correct
decision.
,To further illustrate the subjective nature of rationality, consider a man
with a high discount rate whose finances and health are deteriorating
because he gives little weight to the future. Suppose he is offered a
solution: a pill that would make him value future outcomes more. Although
taking the pill would improve his life and generate retrospective gratitude,
a rational agent must refuse it. The actions the pill would induce are
already available to him, and if he does not choose them, it is because he
does not desire to. Consequently, he has no reason to choose a means -
taking the pill - that would force him to act against his existing
preferences.
When rationality is understood as internal coherence rather than
conformity to reason, it becomes clear that what counts as “rational”
action will vary systematically across social contexts. Henrich et al.
(2010)’s cross-cultural experiments show that formal preferences for risk
aversion and intertemporal choice vary systematically across societies.
WEIRD populations tend to be risk-averse and relatively patient, whereas
many small-scale societies display risk-seeking behaviour and much
steeper future discounting. One Amazonian population of forager-
horticulturalists were found to discount the future 10 times more steeply
than WEIRD people. These findings indicate that formal preferences are
culturally shaped rather than universal features of humanity, undermining
any attempt to treat them as exogenous inputs into rational choice
models.
Preference ordering
Utility maximization involves choosing whichever course of action delivers
the highest utility which, in turn, requires that an individual’s preferences
form a coherent and consistent ranking of all possible outcomes. As such,
rational choice theorists assume complete and transitive preferences;
otherwise, the very idea of selecting the highest-utility option utility
becomes incoherent.
However, empirical evidence challenges the RCT assumptions of stable,
complete, and transitive preferences. Tversky & Simonson (1993)
demonstrate that individuals’ preferences between two options frequently
shift when a third “decoy” option is introduced. For example, one may
prefer a cheap small popcorn to an expensive large popcorn, but prefer
the large to the small once a strategically-priced medium option is
introduced. Such experimental evidence directly violates the RCT
assumption of completeness since, rationally, adding new alternatives
should be entirely irrelevant to pre-existing comparisons. Such behavior
suggests that preferences are not taken from a stable, internal ranking but
are instead constructed in the moment based on the specific context of
,the available options. Other empirical evidence shows that the framing of
a choice significantly alters preferences, even when the underlying
options remain identical. For example, presenting an identical outcome in
terms of potential losses rather than gains reliably shifts people's
rankings, further challenging the idea that actors possess a single, well-
defined, and stable ordering over outcomes. Finally, experimental and
behavioural evidence (e.g. Ainslie, 2001) demonstrates that people
discount future rewards hyperbolically, not exponentially, generating
further preference reversals which challenge the assumption of time-
consistent preferences central to many RCT models. Multi-million-pound
industries e.g. gyms and dieting programmes, depend on these
inconsistent preferences.
In response, defenders may note that in its most minimal form, RCT
requires only that preferences be coherent at the moment of choice,
allowing for a utility-maximizing decision to occur, not that they remain
fixed across time or situations. On this view, apparent preference
reversals can be interpreted as cases of context-dependent utility, where
individuals genuinely value options differently depending on the set of
alternatives or the informational framing presented. This does not
undermine rationality but reflects stable, underlying evaluative principles
that respond to contextual cues. Thus, RCT can formally accommodate
behavioural patterns that might otherwise appear irrational.
Nevertheless, whilst RCT can logically accommodate context-dependent or
time-varying preferences, models used in sociology and economics are
forced to assume temporal and context stability for their empirical work to
function. Without this stable foundation, researchers are prevented from
using best-response reasoning to derive equilibria or confidently make
predictions about future behavior. Instability also makes it impossible to
reliably estimate preferences from observed choices, which are needed for
constructing utility functions that are generalizable across different
situations. Thus, without the assumption of stable preferences, RCT loses
its explanatory and predictive power by rendering its formal modelling
tools unusable on real-world data. Hence, evidence of instability remains a
significant challenge to applied RCT by undermining the core axioms of
preference ordering.
Bounded Rationality
Herbert Simon’s theory of bounded rationality holds that cognitive
limitations, information costs, and time pressures prevent agents from
constructing complete and carefully ordered preference rankings over all
alternatives. This is obviously correct, since humans have limited
computational skills and seriously flawed memories. As such, agents
, satisfice and choose actions that are “good enough” rather than strictly
utility-maximising, turning to heuristics and rules of thumb to guide
choice. Within this framework, phenomena such as framing effects or
preference reversals become predictable consequences of decision-
making under cognitive and informational constraints, even if they remain
normatively suboptimal. The popcorn “decoy” example illustrates this
well: when a medium-priced option is introduced, people typically switch
from preferring a small to a large popcorn, not because their underlying
preferences have collapsed, but because the added option changes how
they heuristically evaluate value-for-money in a complex, time-pressured
environment. On this view, agents are responding rationally to their own
cognitive limitations by minimizing the combined costs of deliberation
time and error, even if the heuristics they rely on occasionally produce
deviations from standard rational-choice predictions.
Beliefs and Procedural Rationality
Traditional or objective rational choice theory understands rational action
as the selection of the objectively utility-maximising option, given fixed
preferences and constraints. However, this implicitly presupposes that
agents possess perfect, complete information and the cognitive capacity
to survey all possible alternatives, accurately foresee their consequences,
and compute the utility-maximising action. These assumptions are
empirically implausible. In real decision contexts, agents routinely face
uncertainty, severe informational constraints, strategic interdependence
where payoffs depend on the actions of others, and limited cognitive
resources. One would have to be omniscient to satisfy the demands of
objective utility-maximization. As a result, rational choice theorists have
shifted from objective utility maximisation to the maximisation of
subjective expected utility (SEU), according to which actions are evaluated
relative to the agent’s beliefs rather than to complete or perfect
information. On this view, rationality does not require acting on true
beliefs, but acting coherently on beliefs that are reasonable to hold given
the agent’s circumstances.
Information-gathering is itself costly, involving both direct costs and
opportunity costs. Where further inquiry is believed to be expensive, time-
consuming, or risky - as in situations of possible imminent danger - it may
be rational, and sometimes the only rational option, to refrain from further
inquiry. Further, this situation is not exceptional but characteristic of
everyday life. Most beliefs must be taken second-hand, since attempting
to verify them all would make ordinary action impossible and prevent us
from enjoying our lives. At some point, the actor must decide, on
subjective grounds, to act on the information they already possess.
Rational Choice Theory models social action as the outcome of purposive,
instrumentally rational agents who hold preferences over outcomes, form
beliefs about how alternative actions map onto those outcomes, and who
operate under constraints (resources, institutions, information etc.). Given
these constraints, preferences, and beliefs about the world, actors select
the action they expect will maximizes their utility.
Preferences
‘Rational’ desires and preferences
The word ‘rationality’ invokes the idea of reason, suggesting agents who
follow the prescriptions of instrumental reason: forming justified beliefs,
endorsing appropriate ends, and choosing effective means to reach them.
However, acting in conformity with reason and acting rationally are two
different things. Conformity with reason is judged from an external
perspective, whereas rationality is judged from the agent's internal
standpoint. An action is rational if it follows coherently from the agent’s
own beliefs and preferences, regardless of whether those preferences are
themselves reasonable or well-justified. Theories of rational action
characteristically assume the agent’s desires and preferences as a fixed
given, and only evaluate how agents act given these preferences.
As a result, preferences that appear unreasonable are not necessarily
irrational. Consider a preference for consuming goods on Wednesdays as
opposed to Thursdays, solely because one prefers that day of the week.
Such a preference violates reason but does nothing to violate rational
choice theory, which places no constraints on the content of preferences.
Only incoherent desires are excluded, such as the wish that everyone earn
more than average.
Desires themselves include both substantive preferences, such as
preferring apples to oranges, and formal preferences, such as one’s
attitude towards risk, uncertainty, and the future. Consider a choice
between a guaranteed £100 or a lottery ticket offering a 0.00001%
chance of winning £1,000,000. Taking the lottery ticket may seem
irrational to most but, in expected-value terms, the two options are
equivalent. A risk-averse agent may rationally prefer the guaranteed
£100, but a risk-seeking agent may rationally prefer the lottery. There is
no uniquely “rational” option independent of the agent’s formal
preferences, because rationality concerns coherence between
preferences, beliefs, and choice - not conformity to an objectively correct
decision.
,To further illustrate the subjective nature of rationality, consider a man
with a high discount rate whose finances and health are deteriorating
because he gives little weight to the future. Suppose he is offered a
solution: a pill that would make him value future outcomes more. Although
taking the pill would improve his life and generate retrospective gratitude,
a rational agent must refuse it. The actions the pill would induce are
already available to him, and if he does not choose them, it is because he
does not desire to. Consequently, he has no reason to choose a means -
taking the pill - that would force him to act against his existing
preferences.
When rationality is understood as internal coherence rather than
conformity to reason, it becomes clear that what counts as “rational”
action will vary systematically across social contexts. Henrich et al.
(2010)’s cross-cultural experiments show that formal preferences for risk
aversion and intertemporal choice vary systematically across societies.
WEIRD populations tend to be risk-averse and relatively patient, whereas
many small-scale societies display risk-seeking behaviour and much
steeper future discounting. One Amazonian population of forager-
horticulturalists were found to discount the future 10 times more steeply
than WEIRD people. These findings indicate that formal preferences are
culturally shaped rather than universal features of humanity, undermining
any attempt to treat them as exogenous inputs into rational choice
models.
Preference ordering
Utility maximization involves choosing whichever course of action delivers
the highest utility which, in turn, requires that an individual’s preferences
form a coherent and consistent ranking of all possible outcomes. As such,
rational choice theorists assume complete and transitive preferences;
otherwise, the very idea of selecting the highest-utility option utility
becomes incoherent.
However, empirical evidence challenges the RCT assumptions of stable,
complete, and transitive preferences. Tversky & Simonson (1993)
demonstrate that individuals’ preferences between two options frequently
shift when a third “decoy” option is introduced. For example, one may
prefer a cheap small popcorn to an expensive large popcorn, but prefer
the large to the small once a strategically-priced medium option is
introduced. Such experimental evidence directly violates the RCT
assumption of completeness since, rationally, adding new alternatives
should be entirely irrelevant to pre-existing comparisons. Such behavior
suggests that preferences are not taken from a stable, internal ranking but
are instead constructed in the moment based on the specific context of
,the available options. Other empirical evidence shows that the framing of
a choice significantly alters preferences, even when the underlying
options remain identical. For example, presenting an identical outcome in
terms of potential losses rather than gains reliably shifts people's
rankings, further challenging the idea that actors possess a single, well-
defined, and stable ordering over outcomes. Finally, experimental and
behavioural evidence (e.g. Ainslie, 2001) demonstrates that people
discount future rewards hyperbolically, not exponentially, generating
further preference reversals which challenge the assumption of time-
consistent preferences central to many RCT models. Multi-million-pound
industries e.g. gyms and dieting programmes, depend on these
inconsistent preferences.
In response, defenders may note that in its most minimal form, RCT
requires only that preferences be coherent at the moment of choice,
allowing for a utility-maximizing decision to occur, not that they remain
fixed across time or situations. On this view, apparent preference
reversals can be interpreted as cases of context-dependent utility, where
individuals genuinely value options differently depending on the set of
alternatives or the informational framing presented. This does not
undermine rationality but reflects stable, underlying evaluative principles
that respond to contextual cues. Thus, RCT can formally accommodate
behavioural patterns that might otherwise appear irrational.
Nevertheless, whilst RCT can logically accommodate context-dependent or
time-varying preferences, models used in sociology and economics are
forced to assume temporal and context stability for their empirical work to
function. Without this stable foundation, researchers are prevented from
using best-response reasoning to derive equilibria or confidently make
predictions about future behavior. Instability also makes it impossible to
reliably estimate preferences from observed choices, which are needed for
constructing utility functions that are generalizable across different
situations. Thus, without the assumption of stable preferences, RCT loses
its explanatory and predictive power by rendering its formal modelling
tools unusable on real-world data. Hence, evidence of instability remains a
significant challenge to applied RCT by undermining the core axioms of
preference ordering.
Bounded Rationality
Herbert Simon’s theory of bounded rationality holds that cognitive
limitations, information costs, and time pressures prevent agents from
constructing complete and carefully ordered preference rankings over all
alternatives. This is obviously correct, since humans have limited
computational skills and seriously flawed memories. As such, agents
, satisfice and choose actions that are “good enough” rather than strictly
utility-maximising, turning to heuristics and rules of thumb to guide
choice. Within this framework, phenomena such as framing effects or
preference reversals become predictable consequences of decision-
making under cognitive and informational constraints, even if they remain
normatively suboptimal. The popcorn “decoy” example illustrates this
well: when a medium-priced option is introduced, people typically switch
from preferring a small to a large popcorn, not because their underlying
preferences have collapsed, but because the added option changes how
they heuristically evaluate value-for-money in a complex, time-pressured
environment. On this view, agents are responding rationally to their own
cognitive limitations by minimizing the combined costs of deliberation
time and error, even if the heuristics they rely on occasionally produce
deviations from standard rational-choice predictions.
Beliefs and Procedural Rationality
Traditional or objective rational choice theory understands rational action
as the selection of the objectively utility-maximising option, given fixed
preferences and constraints. However, this implicitly presupposes that
agents possess perfect, complete information and the cognitive capacity
to survey all possible alternatives, accurately foresee their consequences,
and compute the utility-maximising action. These assumptions are
empirically implausible. In real decision contexts, agents routinely face
uncertainty, severe informational constraints, strategic interdependence
where payoffs depend on the actions of others, and limited cognitive
resources. One would have to be omniscient to satisfy the demands of
objective utility-maximization. As a result, rational choice theorists have
shifted from objective utility maximisation to the maximisation of
subjective expected utility (SEU), according to which actions are evaluated
relative to the agent’s beliefs rather than to complete or perfect
information. On this view, rationality does not require acting on true
beliefs, but acting coherently on beliefs that are reasonable to hold given
the agent’s circumstances.
Information-gathering is itself costly, involving both direct costs and
opportunity costs. Where further inquiry is believed to be expensive, time-
consuming, or risky - as in situations of possible imminent danger - it may
be rational, and sometimes the only rational option, to refrain from further
inquiry. Further, this situation is not exceptional but characteristic of
everyday life. Most beliefs must be taken second-hand, since attempting
to verify them all would make ordinary action impossible and prevent us
from enjoying our lives. At some point, the actor must decide, on
subjective grounds, to act on the information they already possess.
