Information foraging in the unknown patches across the life span.

Abstract

This study used a word search puzzle paradigm to examine the effects of task environment and individual differences in ability on information foraging. Younger and older adults attempted to maximize the number of items found in a set of 4 puzzles in which they were at liberty to search within a puzzle or switch between them. Younger adults demonstrated faster uptake (i.e., number of words found as a function of time) from individual puzzles than older adults but experienced more deceleration of rates during the search. Additionally, older adults switched less often and their switching was less dependent on the uptake rate compared to younger adults. Both younger and older adults stayed longer than was optimal in a patch, older adults were especially likely to persevere suboptimally. Collectively, these results suggest that individuals may differentially optimize information gain through self-regulation of exploration and exploitation.

Full text

Information Foraging in the Unknown Patches across the Life Span
Jessie Chin1 (chin5@illinois.edu), Brennan Payne1 (payne12@illinois.edu),
Andrew Battles2 (battles2@illinois.edu), Wai-Tat Fu3 (wfu@illinois.edu),
Daniel Morrow1 (dgm@illinois.edu), Elizabeth A. L. Stine-Morrow1 (eals@illinois.edu)
1Department of Educational Psychology, 2Department of Electrical & Computer Engineering,
3Department of Computer Science
University of Illinois at Urbana Champaign
405 N Mathews Ave
Urbana, IL 61801 USA
Abstract
This study used a word search puzzle paradigm to examine
the effects of task environment and individual differences in
ability on information foraging. Younger and older adults
attempted to maximize the number of items found in a set of
4 puzzles in which they were at liberty to search within a
puzzle or switch between them. Younger adults demonstrated
faster uptake (i.e., number of words found as a function of
time) from individual puzzles than older adults but
experienced more deceleration of rates during the search.
Additionally, older adults switched less often and their
switching was less dependent on the uptake rate compared to
younger adults. Both younger and older adults stayed longer
than was optimal in a patch, older adults were especially
likely to persevere suboptimally. Collectively, these results
suggest that individuals may differentially optimize
information gain through self-regulation of exploration and
exploitation.
Keywords: Information foraging; information uptake;
cognitive aging; adaptive behavior.
Introduction
Self-regulation of cognition in natural environments
almost always involves alternating phases of exploration,
which entails search in the service of deciding how effort
will be allocated, and exploitation, or task engagement in
which effort is allocated to meet task-specific goals.
Information Foraging (IF) models are predicated on an
analogy between these regulatory processes and the way in
which animals forage for food in the wild. Information
foraging has been used to account for how people search for
information in external environments, such as the WWW
(e.g., Fu & Pirolli, 2007; Payne et al., 2007; Pirolli & Card,
1999) and in memory (Hills et al., 2010, 2012). However,
even though IF presents a compelling metaphor, there is
actually very little empirical research investigating the
alignment between IF principles and how people interact
with the environment to search and make use of information
sources (Metcalfe & Jacobs, 2010). There is also little work
that has examined how individual differences afford or
constrain search in and uptake from information sources. In
this study, we used a simple word search puzzle to explore
these issues.
According to the IF theory (Fu & Pirolli, 2007; Pirolli &
Card, 1999), certain basic properties of animal foraging can
be applied to the way human seek and consume information.
First, food is distributed in the wild in clusters, or “patches,”
that vary in their profitability (i.e., potential yield) and in
their tractability (i.e., how much of an investment of
resources is needed for exploitation; e.g., apples on low
branches or high branches). Resources in the patch are often
finite and unknown to the foragers in advance, though
“scent cues” may provide hints about profitability of the
patch. Second, as patches become depleted, the rate of
uptake decelerates. Third, the forager faces a tradeoff
between gaining nutrients from exploiting a patch and
consuming energy from exploring for food (e.g., to move
among patches). The optimal foraging theory predicts that
animals will stay in a patch until the expected rate of gain
falls below the overall rate of gain, which takes into account
the cost of moving to a new patch (Charnov, 1976; Stephens
& Kreb, 1986). Finally, because food is crucial to survival,
foragers work to maximize their food uptake and rarely
revisit a depleted patch (Stephen, Brown & Ydenberg, 2007;
Stephens & Krebs, 1986).
There are similarities and differences between animal
foraging and human information foraging. For example,
information is often clustered into patches (e.g., particular
forms of print resources, webpages), though units of
information are often hard to quantify in everyday life.
Although information seekers may sometimes find it
difficult to estimate profitability and tractability before
visiting a patch, they may judge the richness or relevance of
information based on their knowledge or expertise. Learners
often selectively allocate their attention to materials as long
as they perceive themselves to be learning, and disengage if
they perceive their rate of learning to decrease below a
threshold (e.g., Metcalfe, 2002; Metcalfe & Kornell, 2005).
While information seekers have been found to adjust their
search behavior to the statistical structures of the task
environments (e.g., Fu & Pirolli, 2007), given the limited
computational capacity and imperfect knowledge of human
beings, the decision to explore a new task or exploit the
current one is often suboptimal due to the biased
representation of the local environment (e.g., Simon, 1956).
For example, Payne, Duggan and Neth (2007) found, in a
series of cognitive foraging experiments, that switch
decisions could not be entirely predicted by the rate of gain
from a patch. Rather, people tended to switch more than
optimal without monitoring the real-time change of
expected gain. Finally, empirical studies show that
information seekers often revisit information patches (e.g.
Payne et al., 2007). In fact, unlike food, information will not
be exhausted after consumption. Therefore, the benefit of
1404

“revisiting a patch” is particularly ecologically important in
information foraging.
Little research has examined adult age differences in
foraging behavior. Aging brings changes in both processing
capacity and knowledge that would likely impact both
uptake rates and exploratory behavior (Beier & Ackerman,
2005). In fact, older information seekers have been found to
adopt different strategies to adapt to the environment. Mata,
Wilke and Czienskowski (2009) showed that older adults
were adaptive to the task characteristics in a fish foraging
task, such as staying longer in one pond while between-
ponds travel time was high. Interestingly, older adults have
been found to search/explore less information but use
simple heuristics or knowledge-driven strategies to achieve
good performance in decision-making or ill-defined
information search tasks (e.g., Chin, Fu & Kannampallil,
2009; Mata & Nunes, 2010). However, older adults’
information uptake behavior in a foraging task has generally
received little attention. To investigate information foraging
behavior in unknown environments, the goals of the current
research were to examine: 1) the effects of task
environments and individual differences on information
uptake (measured as the rate of information gain), and 2) the
effects of task environments and individual differences on
the decisions to switch between sources.
Methods
The word search puzzle paradigm was modified from
previous research (e.g., Chin, Fu & Stine-Morrow, 2011;
Experiment 4 in Payne, Duggen & Neth, 2007). Participants
were asked to maximize the number of items found in a set
of 4 word search puzzles on an iPad. One puzzle was visible
at a time and participants switched between puzzles at
liberty, with a 10-minute limit (See Figure 1).
Participants
Sixty-one participants were recruited from the
community. Four participants (3 young, 1 old) were
excluded due to technical problem or failure to comply with
the instructions. Among remaining 57 participants, 28
young adults (Mean Age = 19.79, SD = 1.23; 19 female)
and 29 old adults (Mean Age = 70.57, SD = 6.33, Range =
62-85; 20 female) were analyzed. All participants had
graduated from high school. There was no age difference in
the frequency of iPad use (t(56)=0.55, p=0.59). Young
adults used computers more often than old adults
(t(56)=2.83, p<.01), and old adults did word puzzles more
often than young adults (t(56)=-2.63, p<.05). Older adults
had better vocabulary than younger adults as measured by
the Advanced Vocabulary Test (Ekstrom et al., 1976)
(t(56)=-4.77, p<.001). On the other hand, younger adults
had better working memory than older adults, as measured
by Reading Span task (Stine & Hindman, 1994) (t(56)=2.87,
p<.01).
Materials
The 4 puzzles, each containing 16 words from a different
semantic category, were presented in three conditions: all
easy, containing mostly high-prototypical category
exemplars in canonical orientations in the puzzle (forward,
down, left-right diagonal); all difficult, containing mostly
low-prototypical exemplars in any orientations; and mixed
(2 easy, 2 difficult). Measurement of exemplar
prototypicality was based on category norms from Van
Overschelde, Rawson, and Dunlosky (2004), in which
prototypicality was indexed as the proportion of participants
generating the word when given the category; there was
significant difference in the mean prototypicality of words
in the easy and hard puzzles (F(1,10)=20.82, p<.001). There
were no differences in the mean log word frequency (Balota
et al., 2007, F(1,10)=0.69, p=.42) or mean word length
(F(1,10)=0.20, p=.66) between items in the easy and hard
puzzles. Thus, given that the words in the easy puzzles were
easier to generate from semantic memory and in a canonical
orientation, they were more likely to “pop-out” than those in
the hard puzzles. While controlling the density of the easy
Figure 1. Layout of the word search puzzle experiments
1405

and hard puzzles, we manipulated the profitability of the
puzzles to see if participants were effective in monitoring
their uptake rates.
The interface for word search puzzle was programmed in
iPad (see Figure 1). Participants first saw the interface with
four colored buttons. Each button referred to a puzzle of
different semantic category. Participants could press any of
the four buttons to start the experiment. When the
participant pressed any of the four buttons,
a countdown timer of 10 minutes started. A word search
puzzle appeared with its category name shown on the top
and bottom of the interface. Participants saw one puzzle at a
time, and used their fingers to swipe the words they found.
The found words were highlighted in different colors and
remained highlighted during the whole session. Participants
could check the number of words they found in each puzzle
on the right corner, but would not know the number of
words remaining in each puzzle. During the experiments,
participants could press the button to switch to the other
puzzles. In the mixed condition, the order of buttons of easy
and hard puzzles was in counterbalanced order. Every
meaningful touch (such as button touch, letter touch) on the
iPad was recorded with time stamps.
Experimental Design
The experiment followed a 2 x 3 mixed factor design with
between-subject variable, age (young vs. old) and within-
subject variable, task condition (all easy vs. mixed vs. all
hard). The order of the three conditions was
counterbalanced across participants.
Procedures
At the beginning of the experiment, participants
completed cognitive measures after the consent process.
Participants then practiced locating words in the puzzles and
switching among puzzles for 20 minutes. After the practice,
participants performed the experimental task. Each
condition took 10 minutes. Participants had been told
explicitly that some puzzles might be easier than others, and
they could go back and forth among four puzzles and decide
how long they want to spend in each puzzle on their own.
After all three conditions, the experimenter briefly
interviewed the participants about their self-observed search
and switch strategies. Participants were debriefed at the end.
Results
A 2 x 3 Repeated Measures Analysis of Variance showed
significant main effects of age and condition on the number
of words found in each condition (Age: F(1,55)=35.37,
p<.001; Condition: F(2,55)=191.78, p<.001). Both younger
and older adults found the most words in the Easy
condition, then the Mixed condition, followed by the Hard
condition. Younger adults found more words than older
adults across all the conditions. The Age by Condition
interaction was not significant (see Table 1). However,
younger and older adults varied in the extent to which they
found words on their first encounter (Bout 1) versus
successive encounters (Bout>1) with the puzzles (Figure 2).
Older adults tended to find most words in their first bout at
the puzzle, while younger adults tended to find relatively
more words in later bouts (i.e., more revisiting), especially
in the hard puzzle (Age: F(1,54)=11.00, p<.005).
Age Differences in Uptake Rates
Mixed-effects modeling was conducted to estimate uptake
rates in the different conditions. Uptake rate was defined as
the cumulative number of words found as a function of time
with data modeled based on 2-sec intervals. As showed in
Figure 2, participants found most words in their first bouts
across different conditions; thus, we modeled the uptake
rates for the first bout only. There were 37,763 observations
in total. Following the growth curve analysis method
(Mirman, Dixon & Magnuson, 2008), we started with the
Table 1. Descriptive statistics of word search
performance
Mean (SD)
Easy
Mixed
Hard
Young
38.93(6.35)
31.86(5.63)
23.39(7.40)
Old
29.24(6.95)
23.34(5.47)
15.72(6.15)
Figure 2. Age difference in the percent of words found in the
first attempt
0!0.2!0.4!0.6!0.8!1!
Easy !
Mixed-easy !
Mixed-hard !
Hard !
Easy !
Mixed-easy !
Mixed-hard !
Hard !
old!
young!
Percentage (%)!
Bout 1!
Bout >1!
Figure 3. The uptake rates for younger and older adults in
different puzzles
− Easy
− Mixed easy
− Mixed hard
− Hard
Young
Old
Time (sec)
Time (sec)
Number of
words
Number of
words
1406

“average uptake rate model.” The uptake rate function
(cumulative number of words per unit time) was calculated
as :
In (1), Yij , γ , U, and eij represented the cumulative number
of words, the fixed effects, the random effects of subjects,
and the error term respectively. Because we are interested in
capturing both the linear and non-linear components of the
random effects of subjects, it was divided into: U1j – the
linear “rate” and U2j – the non-linear “rate of change”. Then
we added fixed effects of age and condition and its
interaction terms to the model “conditional uptake rate
model”, as follow:
The condition update rate model in (2) was developed to
test how uptake rates changed (both linearly and non-
linearly) with conditions and age. The model shows that the
uptake rate (which measured how quickly subjects found a
word in a puzzle) for the easy puzzles was higher than for
the hard ones (F=2377.28, p<.001). Interestingly, the uptake
rate for the hard puzzles was higher when they were
embedded in the mixed condition with easier puzzles
relative to those in the pure condition. This was true for both
younger and older adults, suggesting a facilitation effect in
the mixed condition, in which there were 2 easy and 2 hard
puzzles. Figure 3 showed best fitting curves of uptake rates
of younger and older adults in four puzzles to the empirical
data. The length of curves represents the mean duration of
uptakes (exploitation). As shown in these plots, older adults
stayed longer in the puzzle than younger adults.
Younger adults had higher uptake rates than older adults,
especially in the easy puzzles (Age x condition x time:
F=108.32, p<.001). Younger adults also showed a larger
rate of change, such as quicker deceleration of uptake rate
across time, than older adults (F=16.30, p<.001). The
difference in rates of change was larger in the easy, mixed
puzzles than the hard ones (F=93, p<.001). Thus, the uptake
rates grew more quickly for younger adults but reached the
asymptote quicker (with larger reduction of rates across
time) than older adults.
Age Differences in Switch
Given the individual difference in uptake rates across
different conditions, we examined whether age differences
in uptake rates were related to frequency of switching. A 2 x
3 Repeated Measures Analysis of Variance (Age x
Condition) was conducted on the number of switches in the
easy, mixed, and hard condition. Younger adults switched
more often than older adults in all conditions (Figure 4)
(F(1,55)=30.39, p<.001). There was also a main effect of
condition showing that people switched more in the hard
condition, then the mixed condition, followed by the easy
condition (F(2, 55)=5.21, p<.01). The Age x Condition
interaction was not significant.
Given that younger and older adults experienced different
degrees of rates of change, we examined if the age
differences in rates of change over time were associated
with their switch behavior, and the extent to which they
were moderated by individual differences in working
memory and verbal ability. We first extracted the best linear
unbiased predictors of rates of change from the average
uptake rate model (U2j). Then we did a median split on the
estimates of rates of change to create two groups – those
with uptake rates dropping more and those with uptake rates
dropping less. We did a 2 (Age) x 2 (dropping more or less)
ANCOVA to examine the relationship between the number
of switches and the deceleration of uptake rates across time
by treating individual differences in working memory and
verbal ability as covariates.
Results showed a significant Age x Rate of change
interaction (F(1,51)=7.21, p<.01)) in addition to the effects
of age, rate of change, and working memory (Age:
F(1,51)=19.29, p<.001; Rate of change: F(1,51)=9.87,
p<.01; Working memory: F(1,51)=6.28, p<.05)). The
interaction of Age x Rate of change on the number of
switches was shown in Figure 5. People with more
reduction of uptake rates (the dropping more group) across
time tended to switch more, and the difference was bigger in
younger adults than older adults. In the other words, older
adults were less sensitive to their rates of change in
uptake—they were less likely than younger adults to switch
puzzles as the rate of uptake diminished. On the other hand,
younger adults were more sensitive to changes in uptake
rates, which led to more switches. Also, the covariates of
working memory had association with switch behavior –
people with higher working memory capacity tended to
switch more often. However, the age-differences in
associations between rates of change and switch behavior
were shown regardless of the individual differences in
working memory.
Yij= γ1(time) + γ2(time2) + U1j(time) + U2j (time2) + eij
(1)
Yij= γ1(Time) + γ2(Time2) + γ3(Age x Time) + γ4(Age x
Time2) + γ5(Condition x Time) + γ6(Condition x Time2) +
γ7(Age x Condition x Time) + γ8(Age x Condition x Time2)
+ U1j(time) + U2j (time2) + eij
(2)
Figure 4. Age differences in number of switches of different
conditions
0!
5!
10!
15!
Easy!
Mixed!
Hard!
number of switches!
young!
old!
1407

Suboptimal Leaving and Longer Perseverance for Older
Adults
Given that younger adults switched more than older
adults, and also showed higher uptake rates and rates of
change, the next question we addressed was whether
younger and older participants switched optimally.
According to the optimal foraging theory, the marginal
value theorem predicts the optimal patch departure time –
the time at which the marginal uptake rate is equal to the
mean uptake rate of the entire habitat (Charnov, 1976). We
calculated the ratio of marginal uptake rate at each word and
the mean uptake rate of the corresponding patch for each
participant. The optimal time to switch to a different puzzle
is when the ratio equals 1. When the ratio is larger than 1, it
is advantageous to stay because the current marginal rate is
higher than the average expected return (estimated based on
previous experiences). As the marginal value decreases with
decelerated uptake rates, the value becomes increasingly
smaller than 1, and it is advantageous to switch because the
expected uptake from the habitat as a whole exceeds the
current marginal value.
Both younger and older adults were suboptimal based on
the criterion derived from the marginal value theorem, as
they left the puzzle late (Figure 6a). Mean ratio of the last
word in the puzzle was smaller than 1, suggesting that the
marginal uptake rate of the last word was slower than the
mean uptake rate in the corresponding puzzle. Though
people tended to leave the puzzle when the uptake rate was
low, Figure 6a shows that participants would have been
more optimal if they left the puzzle about 2 words earlier
(the ratio of the third word back was close to 1).
Additionally, among puzzles of different profitability,
people tended to switch more optimally in the hard puzzles
than in the easier puzzles. This finding suggests that both
younger and older adults were more sensitive to the change
of uptake rates and switched earlier in the hard puzzle
condition (Figure 6b).
While both younger and older adults switched later than
was optimal, they persevered differently in the puzzles.
Perseverance was measured by the give up time, which was
defined as the duration from finding the last word to leaving
the puzzle (e.g., Payne et al, 2007) – i.e., the amount of time
participants persevere in a patch without finding a word.
A 2 x 4 Repeated Measures Analysis of Variance (Age x
Puzzle type: easy, mixed easy, mixed hard, hard) was
conducted to explore the effects of age and puzzle type on
give up time. Give up time was longer for older adults than
for younger adults. In other words, older adults persevered
longer in the current patch before moving to a new patch
compared to younger adults. Figure 6a also showed that
while younger adults tended to persevere for a shorter time
than their mean uptake time for the puzzle, older adults
tended to persevere longer than their mean uptake time.
Furthermore, people persevered longer in the hard puzzles
than the easy ones (puzzle type: F(1,53)=2.55, p<.05; age:
F(1,53)=13.15, p<.001). Interestingly, the give up time in
the mixed easy puzzles was relatively longer than the time
in the all easy condition, suggesting that participants were
influenced by the mixed context.
Conclusion
The study used the word search puzzle paradigm to study
the information search behavior of younger and older adults
in the patches of different profitability. Although the gain
functions of puzzles were unknown to the participants,
individuals were able to allocate their effort to uptake and
switch when uptake decreased. Older adults showed slower
Figure 5. Interaction of age and rate of change on switch
0!
10!
20!
30!
40!
50!
Dropping more!
Dropping less!
Number of switches!
Rate of change!
Young!
Old!
Figure 6. The ratio of marginal uptake rate of the word and
mean uptake rate in the corresponding patch for younger and
older adults (a) and different puzzles (b)
0.6!
0.8!
1!
1.2!
1.4!
-3!
-2!
-1!
Give Up!
1!
2!
ratio (1 = optimal)!
Word order!
Marginal uptake rate!
Mean uptake rate in the patch!
Young!
Old!
(b)!
(a)!
0.4!
0.6!
0.8!
1!
1.2!
1.4!
1.6!
-3!
-2!
-1!
Give Up!
ratio (optimal = 1)!
Word Order!
Easy!
Mixed easy!
Mixed hard!
Hard!
1408

uptake rates and smaller change of rates than younger adults
across different puzzles. Thus, older adults relied less on the
deceleration of uptake rates to decide when to switch to a
different puzzle. Older adults switched less often and
persevered longer in the puzzles, especially in the difficult
condition. To maximize the search performance, older
adults allocated more time to exploitation (i.e., task
engagement in the puzzles) and younger adults did more
exploration to the new puzzles than the older adults.
Overall, older and younger adults showed adaptive self-
regulation patterns through differential attention to
exploitation and exploration.
Older adults were found to be less explorative in
information search in decision making (Mata & Nunes,
2010) and web information search (Chin, Fu &
Kannampallil, 2009), and they explored (i.e., switched to
another puzzle) less often in the current study as well. Less
exploration might be adaptive given the heavy demands on
processing capacities of switching behavior in information
search (e.g., Chin et al., 2009, 2011). In addition to the
higher switch cost of older adults, results suggested that
older adults seemed to use different policies (i.e., less
relying on the rates of change) to make switch decision than
younger adults. As the optimal foraging model suggests,
foragers will leave while the marginal uptake rates is lower
than the mean uptake function of a patch. However, given
the uptake function of the puzzles were unknown to the
participants, people needed to track their uptake behavior
after entering a puzzle across time to estimate the expected
gain of the puzzle. This process was so information
intensive and resource demanding that older adults might
experience more difficulty executing which was partly
shown in our results. Thus, age differences in learning from
experiences in a given information patch and the
corresponding patch-leaving policy should be further
examined in future studies.
Despite the age differences in switch, both younger and
older adults were suboptimal in terms of the later departure
time in the patches. Interestingly, past studies also found
that foragers were suboptimal in external search task (e.g.,
Mata et al., 2009), but closer to optimal in memory search
(Hills et al., 2012). Hills and his colleagues used cross-
modal priming to show that external search patterns can be
transferred to internal search patterns, suggesting that there
is a central executive control process monitoring both
internal and external search behavior. Therefore, the
difference of patch-departure behavior in internal and
external search task might be due to the fact that foragers
have more knowledge about the gain function of a patch in
the internal search task than the external search task.
Similarly, in the condition of mixed uptake functions,
results showed that people were farther away from the
optimal (i.e., late departure) in the mixed easy, mixed hard
puzzles than the easy and hard puzzles respectively
suggesting that the knowledge of a patch might be important
to determine the optimal departure in the task.
References
Balota, D. A., Yap, M. J., Cortese, M. J., Hutchison, K.A., Kessler, B.,
Loftis, B., Neely, J.H., Nelson, D.L., Simpson, G.B., & Treiman, R.
(2007). The English Lexicon Project. Behavior Research Methods,
39, 445-459.
Beier, M.E. & Ackerman, P.L. (2005). Age, ability and the role of
prior knowledge on the acquisition of new domain knowledge.
Psychology & Aging, 20, 341-355.
Charnov, E. L., (1976). Optimal foraging, the Marginal Value
Theorem. Theoretical Population Biology, 9, 129–136.
Chin, J., Fu, W-T. & Kannampallil, T. (2009). Adaptive information
search: Age-dependent interactions between cognitive profiles and
strategies. In Proceedings of the 27th ACM Conference on Human
Factors in Computing Systems CHI’09 (pp. 1683-1692). ACM
Press.
Chin, J., Fu, W-T. & Stine-Morrow, E.A.L. (2011). To go or to stay:
Age differences in cognitive foraging. Poster of the 33rd CogSci
Conference, Boston, MA.
Ekstrom, R. B., French, J. W., & Harmon, H. H. (1976). Manual for
the Kit of Factor-Referenced Cognitive Tests. Princeton. NJ:
Educational Testing Service.
Fu, W.-T., & Pirolli, P. (2007). SNIF-ACT: A cognitive model of user
navigation on the World Wide Web. Human-Computer Interaction,
22, 355-412.
Hills, T., Todd, P. M., & Goldstone, R. L. (2010). The central
executive as a search process: Priming exploration and exploitation
across domains. Journal of Experimental Psychology: General, 139,
590-609.
Hills, T., Jones, M., & Todd, P.M. (2012). Optimal foraging in
semantic memory. Psychological review, 119, 431-440.
Mata, R., & Nunes, L. (2010). When less is enough: Cognitive aging,
information search and decision quality in consumer choice.
Psychology and Aging, 25, 289-298.
Mata, R., Wilke, A., & Czienskowski, U. (2009). Cognitive aging and
adaptive foraging behavior. Journal of Gerontology: Psychological
Sciences, 64B, 474-481.
Metcalfe, J. (2002). Is study time allocated selectively to a region of
proximal learning? Journal of Experimental Psychology: General,
131, 349-363.
Metcalfe, J., & Kornell, N. (2005). A region of proximal learning
model of study time allocation. Journal of Memory and Language,
52, 463-477.
Mirman, D., Dixon, J.A.,& Magnuson, J.S.(2008). Statistical and
computational models of visual word paradigm: Growth curves and
individual differences. Journal of Memory and Language, 59. 475-
494.
Payne, S. J., Duggan, G. B., & Neth, H. (2007). Discretionary task
interleaving: Heuristics for time allocation in cognitive foraging.
Journal of Experimental Psychology: General, 136, 370-388.
Pirolli, P., & Card, S. (1999). Information foraging. Psychological
Review, 106, 643-675.
Simon, H. A. (1956). Rational choice and the structure of the
environment. Psychology review, 63, 129-138.
Stine, E. A. L., & Hindman, J. (1994). Age differences in reading time
allocation for propositionally dense sentences. Aging and Cognition,
1, 2–16.
Stephens, D. W., & Krebs, J. R. (1986). Foraging theory. Princeton,
NJ: Princeton University Press.
Stephens, D.W., Brown, J.S., and Ydenberg, R.C. (2007). Foraging:
Behavior and Ecology. Chicago: University of Chicago Press.
Van Overschelde, J. P., Rawson, K. A. & Dunlosky, J. (2004).
Category norms: An updated and expanded version of the Batting
and Montage (1969) norms. Journal of Memory and Language, 50,
289-335.
1409