You are on page 1of 4

198 VOLUME 10, NUMBER 6, DECEMBER 2001

Opportunism in Memory: Preparing for needed resources become available,


and not otherwise. This would re-
Chance Encounters quire that each new cue be com-
pared with all pending goals in
Colleen M. Seifert1 and Andrea L. Patalano memory. To accomplish this, an
Department of Psychology, University of Michigan, Ann Arbor, Michigan (C.M.S.), early computational model of oppor-
and Beckman Institute, University of Illinois, Champaign-Urbana, Illinois (A.L.P.) tunism in planning (Hayes-Roth &
Hayes-Roth, 1979) posited special-
ized goal agents that operate in par-
Abstract intentions to return to pending allel outside of conscious processes.
Recognizing opportunities to goals are generated frequently. At According to this model, these
achieve pending goals is an im- the same time, people are continu- agents watch for and notice con-
portant cognitive ability. But ally confronted with rich environ- ditions relevant to them, and then
when and how do we recog- ments containing potential connec- jump in to interrupt current pro-
nize that a current situation is tions to their goals. What determines cessing to satisfy their goal. Their
especially suited to resuming a whether someone recognizes an op- specific expertise (about conditions
past goal? The predictive encod- portunity to satisfy a pending goal? for specific plans, or how to cluster
ing model suggests pending This problem of opportunism goals) gives the agents the ability
goals are encoded into memory would seem to require elaborate to recognize whether novel circum-
in association with anticipated reasoning: How else can someone stances are relevant to them.
environmental features. Opti- identify one of many pending Birnbaum and Collins (1984) ar-
mally, these features are (a) nec- goals that is particularly relevant in gued that goals themselves can
essary for successful goal the current environment? For ex- monitor information about their rele-
satisfaction, (b) distinctive pre- ample, suppose you have recently vance to the current situation. For
conditions for expecting a plan moved to a new home and need to example, suppose you are both
to achieve the goal, and (c) de- change your address at the motor hungry and thirsty, and you decide
scribed so as to be readily iden- vehicle bureau. You make a mental to go out to a restaurant. If you pass
tified in the environment. Later, note of your intention to go to the a water fountain along the way,
ordinary perception of features in bureau office and fill out the re- you will likely recognize it as an op-
the environment leads to auto- quired forms. But when, if ever, will portunity to satisfy your thirst ahead
matic recognition of opportuni- you return to this pending goal? You of schedule. This suggests some pro-
ties already prepared in memory. may fail to do so, or you may adopt a cess must constantly reason about
Evidence from experimental strategy like creating an external re- the relevance of new information (the
studies supports this theory, minder (e.g., a note on your calendar; water fountain), and insert its goal
and demonstrates that general cf. Ceci & Bronfenbrenner, 1985). But into current actions (stop for a drink).
preparation can produce appar- suppose, days later, you drive down- In fact, Freud (1935) suggested this
ently novel opportunism. These town for a scheduled lunch appoint- same account to explain slips of the
findings suggest ways to facili- ment and find it has been unexpect- tongue, in which unexpressed goals
tate the recognition of opportu- edly canceled, so that you have some interrupt and insert themselves into
nities to satisfy pending goals. free time before your next appoint- planned speech acts (as when a long-
ment. Will you recognize this circum- time bachelor asked, How long
stance as a good time to pursue the have you been buried ? when he
Keywords goal of changing your address? The meant to say, married). However, if
prospective memory; planning; ability to notice this event as an op- goals themselves have these capabili-
opportunism; encoding; retrieval portunity, and to shift your planning ties, no opportunities would ever be
strategies accordingly, may be con- missed.
sidered a hallmark of intelligent goal A simpler alternative handles
In a dynamic world, people are pursuit (Schank & Abelson, 1977). the problem of connecting later cues
sometimes forced to postpone pur- to goals in memory by noticing
suit of their goals. For some goals, these connections at the time of
a plan can be created, but a needed memory encoding. If a pending goal
resource (e.g., time, location, or MODELS OF OPPORTUNISM has already been associated with a
tools) may be currently unavailable. specific cue, the later presence of the
Because many goals are pursued si- Ideally, one would recall a goal cue in the environment automati-
multaneously in everyday life, such from memory whenever the cally brings the goal to mind (this is

Published by Blackwell Publishers Inc.


CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 199

what is called event-based re- vehicle bureau requires being down- (Patalano & Seifert, 1997; Patalano,
trieval; Brandimonte & Passol- town at the bureau office, having sev- Seifert, & Hammond, 1993). Our par-
unghi, 1994). For example, you may eral hours free, having your current ticipants engaged in a commonsense
be reminded that you need to license with you, and being there planning task with multiple goals,
change your address when you during office hours. Once these and then were given a cued-recall
happen to drive by the motor vehi- features have been prepared in test of memory for the goals. Specif-
cle bureau, or when a family mem- memory together with the pending ically, we tested the hypothesis that
ber mentions his or her own need goal, you are ready to notice re- recognition of opportunities to
to register. More specific cues have lated opportunities on the basis of achieve goals is enhanced if those
been found to work better than external cues in the environment. opportunities are anticipated dur-
more general ones for event-based This predictive encoding process in- ing encoding. The method involved
retrieval; for example, a person fill- volves the following steps: a planning scenario familiar to the
ing out a questionnaire will find it college students who were our par-
1. A goal that cannot be fit into
easier to remember to perform a fu- ticipants:
current activity is postponed.
ture task if told to do so after finish-
2. The planner reasons about the Imagine you are visiting your friend,
ing the Black Panthers question
circumstances required for its Chris, in her dormitory room. A neigh-
than after finishing the last survey
satisfaction. bor summons Chris to attend a hall
question (Loftus, 1971). McDaniel,
3. The postponed goal is then en- meeting, and she leaves you alone in
Robinson-Riegler, and Einstein
coded into memory in associa- her room. You decide to snoop around
(1998) have shown that perceiving
tion with these features. the room, and if youre careful to leave
a cue as related to a goal may de- no signs, shell never find out.
4. During other activities, the pre-
pend on the current context; for ex-
dicted features may be per-
ample, studying the word barn Within this scenario, we presented
ceived in the environment.
as a cue may not result in retrieval a series of goals constrained by
5. When these features are perceived,
when a farmyard scene including a common objects, for example:
the goal associated with them is
barn is presented. In their studies,
also brought to mind.
cues were assigned to goals at ran- You notice that Chris left her
6. The planner can then reason
dom (e.g., perform action X when new college ring on her bureau.
about whether to resume pur-
you see cue Y); in more naturalis- You try it on your finger, and it
suit of the pending goal.
tic tasks, cues have functional rela- gets stuck. You need to get the
tionships to goals, and opportun- The success of this approach de- ring off before Chris returns.
ism may require identifying the pends on the ability to predict You jump on the bed. In the pro-
relationship of novel cues to pend- which features are distinctive in suc- cess, you manage to leave scuff
ing goals. cessful goal pursuit. For example, marks high up on the white wall
having your current ID with you is a next to the bed. You need to re-
necessary precondition for changing move the scuff marks before
your address at the motor vehicle Chris returns.
PREDICTIVE ENCODING OF bureau, but having your ID is not an When you open the window to
PENDING GOALS indication of a particularly fortu- get some fresh air, a breeze blows
itous time to resume pursuit of this her poster off the wall. You are
Our own model further develops goal. The more you can anticipate not sure how it was attached to
the idea of event-based retrieval to circumstances predictive of success- the wall, but you need to reattach
suggest how the types of features ful goal pursuit, the better you will it before Chris returns.
considered during encoding deter- be at noticing true opportunities
mine success in recognizing later when they occur. During an initial study phase,
opportunities (Seifert et al., 1994). participants were told to read and
In our account, the same planning make a mental note of each goal. In
processes that first recommend post- addition, we manipulated the type
poning further pursuit of a goal also EVIDENCE FOR of preparation performed during
predict which features may help to PREDICTIVE ENCODING encoding of the goals. Participants
satisfy the goal. First, a plan is cre- in one group were given an object
ated, and special resources, tools, In a series of studies, we investi- with the goal, and were asked to
locations, agents, skills, or times re- gated whether the predictive en- generate their own plan using that
quired are identified. For example, coding model successfully accounts object; for example, for the stuck
changing your address at the motor for the recognition of opportunities ring goal, they saw the cue, You

Copyright 2001 American Psychological Society


200 VOLUME 10, NUMBER 6, DECEMBER 2001

think that if only you had some Vase- relevant goal in memory. All par- ties play a key role in opportunis-
line, you might be able to ___? A ticipants had plenty of time to con- tic remindings (e.g., Vaseline
second group was given both a sider the cue object and recognize and soap share a functional role
plan and an object for each goal; for that it could be helpful for the goal. in the plan of lubricating the finger
example, they were told, You However, those who anticipated the with the stuck ring). Predictive en-
think that if only you had some cue were more likely to report it as coding suggests that the planner
Vaseline, you might be able to an opportunity than those who did must create functional associations
grease your finger and slide the not anticipate it. with possible plans in advance in or-
ring off. These instructional ma- Must you anticipate exactly the der to later recognize specific ob-
nipulations were intended to create object later seen as a cue in order to jects as opportunities for goal satis-
differences in how participants en- recognize it as an opportunity? In faction.
coded the goals into memory. other studies, we found that antici- A further challenge for this the-
Next, a recall test presented a se- pating the plan helped people rec- ory is to explain exactly what con-
ries of cues, and the participants ognize novel objects as opportuni- stitutes a predictive feature. What
were asked to write down any of the ties; for example, both Vaseline level of description is best for pre-
studied goals that came to mind. and butter facilitated recall for dicting opportunities within a do-
Each cue described a single every- participants who had prepared the main? For example, with the ad-
day object (e.g., The only thing you lubrication with Vaseline plan for dress change goal, you might
find under the sink is a jar of Vase- the stuck ring goal. The advan- anticipate location, as in Next time
line. If you could use the Vaseline tage of predictive encoding is ready Im in the vicinity of the motor vehi-
to achieve any of your goals, record recognition of not just the specific ob- cle bureau, Ill change my address.
it below.). The cue presented could jects anticipated, but also other objects However, this specific opportunity
match what was studied during that fit the plan. In sum, the advan- may not arise for some time. At the
encoding (e.g., lubrication with Va- tages in retrieval observed were spe- other extreme, prepared features
seline or Vaseline) or present a cific to the type of plan predicted, but may be too general, as in the plan
novel opportunity (an object in- not to the specific object anticipated Next time I have free time, Ill
volved in an equally plausible plan, with each plan. These results support change my address. This feature
but not studied; e.g., ice cubes as a the hypothesis that predictive encod- may occur more often, but noticing
cue for the plan to shrink your fin- ing of cue features can improve the it may be more difficult. Optimally,
ger with cold ice to remove the likelihood of noticing later opportuni- predicted features are (a) necessary
ring). In addition, each participant ties. circumstances for satisfying the goal,
saw five filler cues that were less (b) selected as distinctive conditions
readily associated with any of the for executing a plan, and (c) formu-
goals (e.g., a comb, tea bags, a lated so as to be readily identified in
shoe). The dependent measure was CONCLUSIONS the environment. For example, antic-
the percentage of goals recalled ipating a canceled appointment as
from the cue (e.g., Vaseline) when These experimental findings just the circumstance needed for
it was anticipated (the participant support the predictive encoding you to pursue the goal of changing
saw Vaseline during planning) or model, in which planning features your address would maximize
unanticipated (but related; the par- are predicted at the time of goal your chance of recognizing this type
ticipant saw ice cubes during postponement, indexed with the of opportunity.
planning). goal in memory, and later brought The ability to generate descrip-
As expected, more goals were to mind by cues experienced in the tions of predictive features may
recalled in response to anticipated environment. Although McDaniel improve with experience within a
cues: If Vaseline was studied et al. (1998) showed cue associations domain. With experience, a plan-
with the goal of removing the stuck at the time of encoding can lead to ning vocabulary of available re-
ring, participants recalled the successful retrieval, the present find- sources and critical constraints may
stuck ring goal given the Vase- ings show opportunism can also be be identified, leading to better antic-
line cue. Participants who studied based on novel cues not present dur- ipation of features that indicate op-
ice cubes with the stuck ring ing encoding. They also go beyond portunities. We expect individuals
goal were less likely to see the goal the results of Tulving and Thomson to vary in their success at recogniz-
as related to the Vaseline re- (1973), who demonstrated that re- ing opportunities, to improve with
trieval cue. Unless prepared for the call is facilitated by similarity in en- experience within a domain, and to
specific opportunity, participants coding and retrieval cues, because be limited by the quality of their
did not connect the later cue to its we found that functional similari- planning (cf. Einstein & McDaniel,

Published by Blackwell Publishers Inc.


CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 201

1990). Predictive encoding requires some opportunities despite our ef- References
high quality in plan preparation. If forts to prepare for them; however,
Birnbaum, L.A., & Collins, G.C. (1984). Opportu-
planning is successful, the impetus through predictive encoding, we can nistic planning and Freudian slips. In Proceed-
to return to postponed goals will maximize the detection of those op- ings of the Sixth Annual Conference of the
Cognitive Science Society (pp. 124127). Hills-
arise automatically from features portunities we expect are most dale, NJ: Erlbaum.
in the environment; but if planning likely to arise. And to the extent Brandimonte, M.A., & Passolunghi, M.C. (1994).
The effect of cue-familiarity, cue-distinctive-
is not successful, one may fail to re- that we can plan ahead to accom- ness, and retention interval on prospective re-
turn to pending goals, and will miss plish our goals in the world, chance membering. Quarterly Journal of Experimental
Psychology, 47A, 565587.
opportunities that occur. And with encounters will favor our plans.
Ceci, S.J., & Bronfenbrenner, U. (1985). Dont for-
many goals active at any given get to take the cake out of the oven: Prospec-
time, the choice of which actions to tive memory, strategic time-monitoring, and
Recommended Reading context. Child Development, 56, 152164.
pursue next may be largely directed Einstein, G.O., & McDaniel, M.A. (1990). Normal
by the events one experiences in the Kvavilashvili, L. (1998). Remember- aging and prospective memory. Journal of Ex-
perimental Psychology: Learning, Memory, and
world (Gollwitzer, 1999; Marsh, ing intentions: Testing a new
Cognition, 16, 717726.
Hicks, & Landau, 1998). Responding method of investigation. Applied
Freud, S. (1935). A general introduction to psycho-
Cognitive Psychology, 12, 533554.
based on observed features may be analysis. New York: Liveright.
Marsh, R.L., Hicks, J.L., & Bink, M.L. Gollwitzer, P.M. (1999). Implementation inten-
especially advantageous in fast- (1998). Activation of completed, tions: Strong effects of simple plans. American
paced, dynamic environments. uncompleted, and partially com- Psychologist, 54, 493503.
Predictive encoding is an impor- pleted intentions. Journal of Experi- Hayes-Roth, B., & Hayes-Roth, F. (1979). A cogni-
mental Psychology: Learning, tive model of planning. Cognitive Science, 3 ,
tant phenomenon because it ties the 275310.
Memory, and Cognition, 24, 350361.
intelligent pursuit of goals to ones Loftus, E.F. (1971). Memory for intentions: The ef-
McDaniel, M.A., Glisky, E.L., Guynn, fect of presence of a cue and interpolated ac-
advance preparation. However, in M.J., & Routhieaux, B.C. (1999). tivity. Psychonomic Science, 23, 315316.
some environments, it is not possi- Prospective memory: A neuropsy- Marsh, R.L., Hicks, J.L., & Landau, J.D. (1998). An
ble to stop and plan ahead while in chological study. Neuropsychol- investigation of everyday prospective mem-
ogy, 13, 103110. ory. Memory & Cognition, 26, 633643.
the midst of other activities. In ad-
McDaniel, M.A., Robinson-Riegler, B., & Einstein,
dition, predictive encoding may be G.O. (1998). Prospective remembering: Percep-
costly when the effort expended in tually driven or conceptually driven pro-
Acknowledgments We thank Kristian cesses? Memory & Cognition, 26, 121134.
advance planning later proves un- Hammond, Travis Seymour, Chris Willis, Patalano, A.L., & Seifert, C.M. (1997). Opportun-
necessary. For some less important Michelle Strasser, Gordon Gibbings, and ism in planning. Cognitive Psychology, 34, 136.
goals, returning at a later time may Barbara Yates for their assistance with Patalano, A.L., Seifert, C.M., & Hammond, K.J.
this research. (1993). Predictive encoding: Planning for op-
not warrant the cost of preparation; portunities. In Proceedings of the Fifteenth An-
for others, opportunities may be so nual Conference of the Cognitive Science Society
(pp. 800805). Hillsdale, NJ: Erlbaum.
readily available, or well learned, Schank, R., & Abelson, R. (1977). Scripts, plans, goals
that predictive encoding is unnec- Note and understanding. Hillsdale, NJ: Erlbaum.
essary. But in the appropriate cir- Seifert, C.M., Hammond, K.J., Johnson, H.M., Con-
1. Address correspondence to Col- verse, T.M., MacDougal, T., & VanderStoep,
cumstances, predictive encoding may S.W. (1994). Case-based learning: Predictive fea-
leen Seifert, Department of Psychol-
represent a means of accomplishing ogy, University of Michigan, 525 East
tures in indexing. Machine Learning, 16, 3756.
Tulving, E., & Thomson, D.M. (1973). Encoding
intelligent planning within dynamic University, Ann Arbor, MI 48109-1109; specificity and retrieval processes in episodic
environments. Of course, we all miss e-mail: seifert@umich.edu. memory. Psychological Review, 80, 359380.

Copyright 2001 American Psychological Society

You might also like