Don't stop thinking about tomorrow: Individual differences in future self-continuity account for saving

Judgment and Decision Making, Vol. 4, No. 4, June 2009, pp. 280–286
Don’t stop thinking about tomorrow: Individual differences in
future self-continuity account for saving
Hal Ersner-Hersh?eld?, M. Tess Garton, Kacey Ballard,
Gregory R. Samanez-Larkin, and Brian Knutson
Department of Psychology
Stanford University
Abstract
Some people ?nd it more dif?cult to delay rewards than others. In three experiments, we tested a “future self-
continuity” hypothesis that individual differences in the perception of one’s present self as continuous with a future self
would be associated with measures of saving in the laboratory and everyday life. Higher future self-continuity (assessed
by a novel index) predicted reduced discounting of future rewards in a laboratory task, more matches in adjectival
descriptions of present and future selves, and greater lifetime accumulation of ?nancial assets (even after controlling for
age and education). In addition to demonstrating the reliability and validity of the future self-continuity index, these
?ndings are consistent with the notion that increased future self-continuity might promote saving for the future.
Keywords: Intertemporal choice, temporal discounting, future self-continuity, saving behavior.
1 Introduction
& Knutson, 2009). Accordingly, neural indices of fu-
ture self-continuity predict valuation of future rewards as-
Self-continuity represents one of the most enduring puz-
sessed one week later (Ersner-Hersh?eld et al., 2009). In
zles of personal identity, pondered by ancient Greek and
three experiments, we further explored this future self-
Buddhist thinkers alike (Conze, 1959; Plutarch, 1932). If
continuity hypothesis by devising an index of individ-
the body changes over time — eventually being replaced
ual differences in future self-continuity and examining
by entirely new material — at what point does the old
whether it would correlate with valuation of delayed re-
self become an entirely new and distinct being? Self-
wards, both in laboratory tasks and with respect to real-
continuity might vary by individual, with some individ-
world savings.
uals endorsing greater connection to their future selves
than others.
Beyond philosophical appeal, individual differences
2 Study 1: Future self-continuity
in the experience of self-continuity could have prag-
matic consequences for ?nancial well-being. Accord-
and temporal discounting
ing to one philosophical account (Par?t, 1971), if indi-
viduals consider their future selves as different people,
We devised a novel psychometric measure of future self-
they may have no more reason to reward the future self
continuity, based on an extension of an existing measure
than to give resources to strangers (Ainslie, 1975; Elster,
of self- vs. other-connectedness (Aron, Aron, & Smollan,
1977; Par?t, 1971, 1987; Pronin, Olivola, & Kennedy,
1992)1, and we sought to establish its test-retest reliabil-
2008; Schelling, 1984; Strotz, 1956; Thaler & Shefrin,
ity and validity in predicting valuation of future rewards.
1981). A continuous variant of this account called the
The measure assessed an individual’s endorsement of
“future self-continuity” hypothesis predicts that people
similarity between present and future selves, as well as
who experience no continuity with a future self should
other potential dimensions of future self-continuity re-
not save for that future self (Ersner-Hersh?eld, Wimmer,
lated to connectedness, liking, and caring for the future
?The authors thank Jeffrey C. Cooper, Jonathan Baron, members of
1The Aron et al. (1992) scale ranges from circles with no overlap to
the SPAN lab, and two anonymous reviewers for helpful comments on
circles with nearly complete overlap. The original scale does not, how-
the manuscript. This research was funded in part by Center on Advanc-
ever, include an “identity” option with completely overlapping circles.
ing Decision Making in Aging Grant AG024597, National Institute on
Though we suspect that a complete identity option would not greatly al-
Aging Grant AG030778, and a FINRA Investor Education Foundation
ter participant responses based on their roughly symmetric distribution,
Award.
future research might include such an option.
280

---

Judgment and Decision Making, Vol. 4, No. 4, June 2009
Future self-continuity and saving
281
Current
Future
Current
Future
Current
Future
Current
Future
Self
Self
Self
Self
Self
Self
Self
Self
Current
Future
Current
Future
Current
Future
Self
Self
Self
Self
Self
Self
Figure 1: Future self-continuity scale.
self. To assess valuation of future versus present rewards,
Temporal Discounting Task. The temporal discounting
we used a temporal discounting task (Frederick, Loewen-
task consisted of 21 choice trials (Kirby & Marakovic,
stein, & O’Donoghue, 2003; Laibson, Repetto, & Tobac-
1996). Each trial included one smaller immediate re-
man, 1998; Mitchell, 1999; Schelling, 1982). We pre-
ward paired with one larger delayed reward. Immediate
dicted that individuals who endorsed greater future self-
values ranged from $15–$83, while the delayed values
continuity would place a higher value on future rewards.2
ranged from $30–$85 over delays of 10–75 days. Choices
were hypothetical. The number of delayed choices was
2.1 Method
counted to index discount rate (Magen, Dweck, & Gross,
2008). While the Kirby and Marakovic (1996) procedure
2.1.1 Participants
excludes data from individuals who either chose all of the
immediate or all of the delayed options, the Magen et al.
Undergraduates from the Stanford University Introduc-
(2008) procedure has the advantage of retaining these in-
tory Psychology pool participated for course credit. One
dividuals for subsequent analysis.
hundred sixty-four individuals (66% female; M = 19.34
years) participated in Wave 1, and ninety-three individu-
als (63% female; M = 19.32 years) participated in Wave
2.1.3 Procedure
2. Part of the Wave 2 sample (n = 65) also participated in
Wave 1; and so test-retest reliability was assessed in these
In a ?rst wave of questionnaires, participants completed
overlapping individuals.
the future self-continuity measure. In the second wave,
participants completed the future self-continuity measure
as well as the hypothetical discounting task (Kirby &
2.1.2 Measures
Maracovic, 1996). Sixty-?ve individuals participated in
Future Self-Continuity Measure. The index of future
both Waves 1 and 2 and thus, completed the future self-
self-continuity featured two questions on a 7-point scale
continuity measure twice (separated by a minimum of
marked at each point by two circles that ranged from de-
one week).
picting no overlap to depicting almost complete overlap
(Figure 1). Participants selected the circle pair that best
2.2 Results and Discussion
described how similar and how connected they felt to a
future self ten years from now. They also rated how much
2.2.1 Reliability
they cared about and liked their future self ten years from
now on 7-point Likert scales. To facilitate comprehen-
To assess test-retest reliability, we correlated future self-
sion, both the “caring” and “liking” items included ver-
continuity indices from Waves 1 and 2. Test-retest reli-
bal anchors at each of the seven points (e.g., the “car-
ability for similarity was high (r63 = .66, p < .001; ? =
ing” scale ranged from “don’t care at all” to “completely
.79). Test-retest reliability was also high for connected-
care”).
ness (r63 = .66, p < .001; ? = .80), caring (r63 = .45, p <
.01; ? = .62), and liking items (r63 = .64, p < .001; ? =
2The only previous study of the relationship between future self-
.78; see Table 1).
continuity and temporal discounting (Frederick, 1999) used a single
item measure (i.e., “How similar/connected are you to your past and
future selves?” for 5-, 10-, 15-, and 20-year intervals on a 1–100
2.2.2 Validity
scale). Correlation of this measure with hypothetical temporal discount-
ing rates did not reveal a signi?cant association. However, the reliability
and validity of this single-item measure of future self-continuity were
Consistent with the prediction that future self-continuity
not reported.
would promote valuation of future rewards, individual

---

Judgment and Decision Making, Vol. 4, No. 4, June 2009
Future self-continuity and saving
282
Table 1: Zero-order correlation matrix of all self-continuity scale items, and number of later choices on the tempo-
ral discounting task. Numbers in parentheses refer to the data wave. “Aggregate” refers to the mean of similarity,
connectedness, caring, and liking from Wave 1.
Variable
1
2
3
4
5
6
7
8
9
10
1. Similar (1)
–
0.47**
0.14
0.16*
0.66**
0.41**
0.22
0.32**
0.75**
0.42**
2. Connect (1)
–
0.20**
0.25**
0.56**
0.66**
0.31*
0.52**
0.82**
0.30*
3. Care (1)
–
0.35**
0.20
0.26*
0.45**
0.33**
0.47**
0.23
4. Like (1)
–
0.36**
0.47**
0.31*
0.64**
0.57**
0.17
5. Similar (2)
–
0.46**
0.13
0.23*
0.67**
0.37**
6. Connect (2)
–
0.37**
0.38**
0.66**
0.07
7. Care (2)
–
0.50**
0.41*
0.21*
8. Like (2)
–
0.61**
0.29**
9. Aggregate
–
0.39**
10. Later choices
–
* p < .05; ** p < .01, 2-tailed
differences in future self similarity (at Wave 1) signi?-
3 Study 2: Future self-continuity
cantly predicted the number of delayed choices at Wave
and self-descriptive consistency
2 (r63 = .42, p < .001; see Figure 2a). Future self connect-
edness (at Wave 1) also predicted the number of delayed
In Study 2, we sought to associate future self-continuity
choices at Wave 2 (r63 = .30, p < .05). Future self caring
with performance on a trait-rating task (called the
and future self liking (at Wave 1), however, did not pre-
“Me/Not Me” task) in which participants endorsed pos-
dict the number of delayed choices at Wave 2 (caring: r63
itive, neutral, and negative trait words for both their cur-
= .21, p = .11; liking: r63 = .14, p = .27).
rent and future selves (Aron, Aron, Tudor, & Nelson,
Although these ?ndings supported the prediction that
1991; Wakslak, Nussbaum, Liberman, & Trope, 2008).
individuals who endorse greater future self-continuity
We predicted that individuals with higher future self-
choose more delayed rewards in a temporal discounting
continuity would show more matches in their endorse-
task, future self-similarity most robustly predicted num-
ment or rejection of current and future self-descriptors
ber of delayed choices and showed high test-retest relia-
in the context of the Me/Not Me task. We also sought
bility. Thus, in two additional studies, we operationalized
to replicate the correlation between future self-continuity
future self-continuity with a measure of perceived simi-
and valuation of future rewards by using an incentive-
larity to the future self.3 In Study 2, we sought to further
compatible temporal discounting task (i.e., in which par-
establish the validity of the future self-continuity index
ticipants received actual money for a subset of their
by examining its association with a distinct and more im-
choices).
plicit measure of future self-continuity, as well as with an
incentive compatible measure of temporal discounting.
3.1 Method
3We also explored the option of creating an aggregate measure of
future self-continuity by averaging similarity, connectedness, caring,
3.1.1 Participants
and liking scores. The aggregate had a high test-retest reliability (r63
= .79, p < .001, ? = .88), but the within-test reliability was not im-
Forty individuals (47.5% female, M = 21.30 years) from
pressive (? = .57). Therefore, we adopted the conservative strategy of
the Stanford University community received a ?at fee of
testing the relationship between each of the four indices of future self-
$15 plus the monetary outcome of one randomly selected
continuity and temporal discounting at a Bonferroni-corrected thresh-
choice in the temporal discounting task (see below) for
old (p = .0125). Of the indices, only future self-similarity survived the
threshold. Moreover, the correlation between the aggregate measure of
their participation.
future self-continuity and number of later choices on the temporal dis-
counting task was not as high as for the one-item similarity measure (r63
= .39, p < .001). Finally, the aggregate did not perform as consistently
3.1.2 Procedure
in predicting saving behavior across the three studies as did the one-
item similarity measure. Thus, we used self-similarity to index future
After signing the consent form, participants read the fol-
self-continuity.
lowing instructions:

---

Judgment and Decision Making, Vol. 4, No. 4, June 2009
Future self-continuity and saving
283
a
b
c
Study 1
Study 2
Study 3
s
s 25
e 100
18
i
ce
t
ch
o 20
90
15
12

ch 15

ma
80
t
s
d
9
i
ve
ye 10
70
ct
Asse
6
l
a
e
j
e
5
d
60
3

d
#

a
0
50
0
%
1
2
3
4
5
6
7
1
2
3
4
5
6
7
1
2
3
4
5
6
7
Future self-similarity
Future self-similarity
Future self-similarity
Figure 2: (a) Endorsement of future self-similarity positively correlates with number of choices of future (versus
present) rewards (Study 1). (b) Endorsement of future self-similarity positively correlates with number of trait matches
on the Me/Not Me task (Study 2). (c) Endorsement of future self-similarity positively correlates with assets (Study 3).
In all three panels, circle size is weighted by number of respondents at each data point.
“In this study you will see a series of trait words. Some
choice (18%) for that trial in cash. If participants had
of the words will apply to you now, and some will apply
chosen the delayed gain for the randomly selected trial,
to you ten years from now. In this part of the study, we are
they received a dated receipt for that amount, which they
interested in how you view yourself now. We will show
were mailed on the appropriate date. Otherwise, if partic-
you a series of words, and we will ask you to indicate
ipants had chosen the immediate gain, they received cash
whether or not each word describes you now.”
corresponding to that gain.
The participant was then instructed to choose “yes” if
the word described the speci?ed person, and “no” if it
did not by pressing correspondingly marked keys. After
3.2 Results and Discussion
completing six practice trials, participants rated whether
96 randomly presented trait words applied to their cur-
We ?rst examined whether future self similarity was as-
rent self, and the words remained on the screen until the
sociated with the number of matches in current and fu-
participant made a choice. The trait words were selected
ture self description (or number of trials in which trait
from the Anderson (1968) word list, which was normal-
adjective ratings were the same for current and future
ized for valence and word length, with one third of the
selves). Consistent with prediction, similarity ratings on
words rated as positive, one third rated as negative, and
the future self-continuity scale correlated with percentage
one third rated as neutral, on average. After rating the ap-
of matches between current and future self-descriptions
plicability of these 96 trait words to the current self, par-
(r38 = .34, p < .05; see Figure 2b). Further, replicat-
ticipants rated the applicability of the same 96 trait words
ing the ?ndings of Experiment 1, higher levels of future
(again in a randomized order) to a future self (10 years
self-similarity correlated positively with the number of
hence), and were given the following instructions:
delayed rewards chosen (r38 = .30, p < .05).
“In this part of the study, we are interested in how you
view yourself in the future (ten years from now). We will
show you a series of words, and we will ask you to indi-
4 Study 3: Future self-continuity
cate whether or not each word describes you in the future
and actual savings
(ten years from now).”
Counterbalanced presentation ensured that half of the
The ?ndings of Studies 1 and 2 suggested that future self-
participants rated their current self ?rst, while the other
continuity is associated with valuing future rewards, as
half rated their future self ?rst. After completing the
indicated by low rates of temporal discounting. Extended
Me/Not Me task, participants completed a question-
over time, low rates of temporal discounting may mani-
naire with demographic information as well as the self-
fest as higher rates of saving, and have been theoretically
continuity scale from Study 1. Finally, participants com-
linked to peoples’ tendency to save for the future (Dia-
pleted an incentive compatible version of the temporal
mond & Koszegi, 2003). The ?ndings of Studies 1 and 2,
discounting task, in which they were informed that one
however, did not establish whether future self-continuity
trial would be randomly selected at the end of the task,
might be associated with actual savings behavior. Thus,
and that they would receive a set percentage of their
in Study 3, we examined the association between future

---

Judgment and Decision Making, Vol. 4, No. 4, June 2009
Future self-continuity and saving
284
self-continuity and the amount of assets that individuals
5 General discussion
had accrued.4
In three studies, we found that individual differences in
4.1 Method
future self-continuity predicted valuation of future re-
wards in laboratory tasks, correlated with the match be-
4.1.1 Participants
tween present and future self-descriptions, and were as-
sociated with real world assets. Together, these ?ndings
155 adult community members from the San Francisco
indicate not only that individuals reliably differ in future
Bay Area participated (45% Female; Age range: 20 —
self-continuity, but also that these individual differences
86 yrs; M = 53.49 yrs).5 These individuals were recruited
may in?uence valuation of future rewards.
as part of a larger study on psychological factors that in-
By establishing the reliability and validity of a mea-
?uence ?nancial decision-making.
sure of self-continuity, the present research could estab-
lish initial evidence for a connection between individual
4.1.2 Procedure
differences in future self-continuity and valuation of fu-
All participants completed a packet of questionnaires
ture rewards. While a previous study did not report a sig-
that assessed self-reported assets (i.e., portion of home
ni?cant association between these variables (Frederick,
owned, total amount of money in bank accounts, total
1999), the pictorial index used in the present study (based
amount of money in investments, and worth of other ma-
on Aron et al., 1992) may have provided individuals with
terial belongings), debts (i.e., outstanding home, car, and
a more tangible and concrete method of reporting their
student loans, credit card debts, and medical debts), in-
perceived future self-continuity. The relation of future
come, and other ?nancial information. To facilitate the
self-continuity to temporal discounting is consistent with
anonymity of responses, assets, debt, and income were
research indicating that neural discrimination of material
rated categorically (16 categories: “$0-$500”,. . . ,“More
related to the present versus future self predicts temporal
than $1,500,000”). The future self-continuity question-
discounting in a subsequent laboratory task (Ersner- Her-
naire from Studies 1 and 2 was also administered. Based
sh?eld et al., 2009). The present study, however, utilized
on previous theoretical work linking discounting to sav-
a simple, quick, and easy psychometric measure of future
ing, we predicted that future self-similarity would corre-
self-continuity rather than neural activity. Unexpectedly,
late with the amount of accrued assets.
future self similarity showed a more robust relation than
future self connectedness to temporal discounting, so an
improved measure might include multiple indices of fu-
4.2 Results
ture self similarity. Nonetheless, the single index of fu-
As predicted, endorsement of future self similarity was
ture self similarity showed high test-retest reliability. Be-
positively associated with assets (r
yond students in the laboratory, the association between
147 = .34, p < .001),
suggesting that the more similar an individual felt to his
individual differences in future self-continuity and choice
or her future self, the more assets he or she had ac-
of delayed rewards further generalized to real-world as-
crued (Figure 2c).6 Although previous research suggests
sets in a community sample.
that age is also positively correlated with saving behavior
Overall, we predicted that increased future self-
(Lea, Tarpy, & Webley, 1987), the association remained
continuity would promote saving for the future self,
signi?cant even after statistically controlling for age (par-
which requires people to trade present gains for future
tial r
gains. However, the extent to which increased future self-
146 = .23, p < .01).7
continuity might also promote loss avoidance has not yet
4Here, we used ?nancial assets as a proxy for saving behavior. Pre-
been systematically explored. Future self-continuity may
vious research has indicated that inclusion of non-?nancial assets (e.g.,
promote valuation of future gains because people might
home ownership) with ?nancial assets (e.g., bank accounts and invest-
ments) may provide a more complete picture of an individual’s saving
?nd it both easier and more pleasant to imagine how their
behavior (e.g., Rha, Montalto, & Hanna, 2006).
actions might lead to positive (rather than negative) future
5Six participants failed to report their income or assets and were thus
outcomes for a future self. Indeed, in a post-hoc analy-
excluded from subsequent analyses.
6
sis of the savings data from Study 3, we found that per-
To rule out the possibility of a spurious relationship resulting from
a correlation between response scale bias (e.g., responding on the right,
ceived similarity to the future self did not correlate sig-
left, or middle of the scale) and assets, we standardized (within-person)
ni?cantly with debts accrued (r147 = .08, p = .36). Still,
the self-continuity scale scores (i.e., by subtracting the mean across
further research is needed to better establish the relation-
items from each item, and dividing this difference by the standard de-
ships between future self-continuity, promotion of posi-
viation) and reran the regression. Even after this transformation, the
relationship between similarity and assets remained signi?cant (r
tive outcomes (e.g., savings), and avoidance of negative
147 =
.29, p < .001).
outcomes (e.g., debt).
7This relationship holds even after controlling for education as well,
which represents another demographic variable with potential relevance
to the study of saving behavior (r145 = .22, p < .01).

---

Judgment and Decision Making, Vol. 4, No. 4, June 2009
Future self-continuity and saving
285
Although the present ?ndings establish a new measure,
individuals who see their future selves as similar may be
they cannot specify the causal direction of the associa-
more likely to sacri?ce present pleasure for the bene?t of
tion between future self-continuity and saving. Consis-
that potential person.
tent with the analysis by Par?t (1987), individuals who
felt more similar to their future selves may have made
more prudent decisions because of their perceived con-
References
nectedness to a future self. Alternatively, anticipated ?-
nancial security and stability might promote future self-
Ainslie, G. (1975). Specious reward: Behavioral theory
continuity such that individuals who imagine being well
of impulsiveness and impulse control. Psychological
off in ten years prefer to identify with that ?nancially se-
Bulletin, 82(4), 463–496.
cure future self. Only experimental manipulations of fu-
Ainslie, G. & Haslam, N. (1992). “Hyperbolic discount-
ture self-continuity can address the causality of the ob-
ing,” in Choice Over Time, ed. G. Loewenstein and J.
served associations, and the measure introduced here of-
Elster, New York, NY: Russell Sage Foundation, 57–
fers one tool for assessing the impact of such manipula-
92.
tions. If future self-continuity in?uences saving (as hy-
Anderson, N. H. (1968). Likableness ratings of 555
pothesized), then interventions that enhance future self-
personality-trait words. Journal of Personality and So-
continuity may promote saving. Future research, perhaps
cial Psychology, 9, 272–279.
in the context of a longitudinal study, might tease apart
Aron, A., Aron, E. N., & Smollan, D. (1992). Inclusion
the causal relationship between future self-continuity and
of other in the self scale and the structure of interper-
saving (see, for example Bartels & Rips, submitted).
sonal closeness. Journal of Personality and Social Psy-
Based on methodological considerations, the timescale
chology, 63, 596–612.
adopted for the self-continuity measure (i.e., 10 years)
Aron, A., Aron, E. N., Tudor, M., & Nelson, G. (1991).
differed from that used in the temporal discounting tasks
Close relationships as including other in the self. Jour-
(i.e., 10–75 days). These timescales were primarily se-
nal of Personality and Social Psychology, 60, 243–253.
lected to match methods used in prior research on future
Bartels, D. M., & Rips, L. J. (submitted). Psychological
self-continuity (Ersner-Hersh?eld et al., 2009) and on
connectedness and temporal discounting.
temporal discounting (Kirby and Maracovic, 1996). Be-
Conze, E. (1959). Buddhist scriptures. Baltimore, MD:
yond precedent, matching timescales might pose method-
Penguin Books.
ological problems, since a self-continuity scale with a
Diamond, P., & Koszegi, B. (2003). Quasi-hyperbolic
shorter timescale might yield ceiling effects (i.e., with
discounting and retirement. Journal of Public Eco-
most subjects endorsing high degrees of similarity with
nomics, 87, 1839–1872.
future selves) and a temporal discounting task with a long
Elster, J. (1977). Ulysses and Sirens: A theory of imper-
timescale might conversely yield ?oor effects (i.e., with
fect rationality. Social Science Information, 16, 469–
subjects discounting most future rewards). Longitudinal
526.
research, however, suggests some invariance in measures
Ersner-Hersh?eld, H., Wimmer, G. E., & Knutson, B.
of temporal discounting (Ainslie & Haslam, 1992), sug-
(2009). Saving for the future self: Neural measures
gesting for instance that pre-schoolers’ ability to delay
of future self-continuity predict temporal discounting.
grati?cation on the order of seconds predicted major de-
Social Cognitive and Affective Neuroscience, 4, 85–92.
velopmental outcomes assessed several years later (e.g.,
Frederick, S. (1999). Discounting, time preference, and
SAT scores, parental ratings of attention, competency,
identity: A dissertation submitted to the Department of
and intelligence; Mischel, Shoda, and Rodriguez (1989)).
Social and Decision Sciences. Carnegie Mellon Uni-
Nonetheless, use of similar timescales might strengthen
versity.
the ?ndings, and future studies will need to investigate
Frederick, S., Loewenstein, G., & O’Donoghue, T.
parametric variations of the timescale used in both self-
(2003). Time discounting and time preference: A
continuity and temporal discounting domains.
critical review. In G. Loewenstein, D. Read & R. F.
Together, these ?ndings provide initial evidence that
Baumeister (Eds.), Time and decision: Economic and
individual differences in perceived future self-continuity
psychological perspectives on intertemporal choice
can promote valuation of future rewards. Speci?cally, the
(pp. 13–86). New York: Russell Sage Foundation.
extent to which individuals perceive the present self as
Kirby, K. N., & Marakovic, N. (1996). Modeling myopic
similar to a future self is associated with less discounting
decisions: Evidence for hyperbolic delay-discounting
of future rewards in the laboratory and greater ?nancial
within subjects and amounts. Organizational Behavior
assets in the real world. From a pragmatic standpoint,
and Human Decision Processes, 64, 22–30.
enhancing future self-continuity might encourage people
Laibson, D., Repetto, A., & Tobacman, J. (1998). Self-
to save for the future. From a philosophical standpoint,
control and saving for retirement. Brookings Papers on

---

Judgment and Decision Making, Vol. 4, No. 4, June 2009
Future self-continuity and saving
286
Economic Activity, 91–196.
Rha, J.-Y., Montalto, C. P., & Hanna, S. (2006). The ef-
Lea, S. E. G., Tarpy, R. M., & Webley, P. (1987). The in-
fect of self-control mechanisms on household saving
dividual in the economy: A textbook of economic psy-
behavior. Financial Counseling and Planning, 17, 3–
chology. New York, NY: Cambridge University Press.
16.
Magen, E., Dweck, C. S., & Gross, J. J. (2008). The
Schelling, T. C. (1982). Ethics, law, and the exercise of
hidden zero effect: Representing a single choice as an
self- command. Boston, MA: Harvard Institute of Eco-
extended sequence reduces impulsive choice. Psycho-
nomic Research.
logical Science, 19, 648–649.
Schelling, T. C. (1984). Self-command in practice, in
Mitchell, S. H. (1999).
Measures of impulsivity in
policy, and in a theory of rational choice. American
cigarette smokers and non-smokers. Psychopharma-
Economic Review, 74, 1–11.
cology (Berl), 146, 455–464.
Strotz, R. H. (1956). Myopia and inconsistency in dy-
Mischel, W., Shoda, Y., & Rodriguez, M. L. (1989). De-
namic utility maximization. Review of Economic Stud-
lay of grati?cation in children. Science, 244, 933–938.
ies, 23, 165–180.
Par?t, D. (1971). Personal Identity. Philosophical Re-
Thaler, R. H., & Shefrin, H. M. (1981). An economic
view, 80, 3–27.
theory of self-control. Journal of Political Economy,
Par?t, D. (1987). Reasons and persons (Repr. with cor-
89, 392–406.
rections. ed.). Oxford: Clarendon Press.
Wakslak, C. J., Nussbaum, S., Liberman, N., & Trope,
Plutarch. (1932). The lives of the noble Grecians and
Y. (2008). Representations of the self in the near and
Romans (J. Dryden, Trans.). New York: The Modern
distant future. Journal of Personality and Social Psy-
Library.
chology, 95, 757–773.
Pronin, E., Olivola, C. Y., & Kennedy, K. A. (2008). Do-
ing unto future selves as you would do unto others:
Psychological distance and decision making. Person-
ality and Social Psychology Bulletin, 34, 224–236.

---