Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Attention deficits linked with proclivity to

royalsocietypublishing.org/journal/rspb
explore while foraging
David L. Barack1,2,†, Vera U. Ludwig1,5,†, Felipe Parodi1, Nuwar Ahmed3,
Elizabeth M. Brannon3, Arjun Ramakrishnan6 and Michael L. Platt1,3,4,5
Research 1
Department of Neuroscience, Perelman School of Medicine, 2Department of Philosophy, 3Department of
Psychology, 4Department of Marketing, Wharton School, and 5University of Pennsylvania, PA 19104, USA
6
Cite this article: Barack DL, Ludwig VU, Department of Biological Sciences and Bioengineering and Mehta Family Centre for Engineering in Medicine,
Parodi F, Ahmed N, Brannon EM, Indian Institute of Technology, Kanpur 208016, India
Ramakrishnan A, Platt ML. 2024 Attention DLB, 0000-0001-7131-4638; MLP, 0000-0003-3912-8821
deficits linked with proclivity to explore while
All mobile organisms forage for resources, choosing how and when to
foraging. Proc. R. Soc. B 291: 20222584.
search for new opportunities by comparing current returns with the average
https://doi.org/10.1098/rspb.2022.2584
Downloaded from https://royalsocietypublishing.org/ on 01 March 2024

for the environment. In humans, nomadic lifestyles favouring exploration


have been associated with genetic mutations implicated in attention deficit
hyperactivity disorder (ADHD), inviting the hypothesis that this condition
may impact foraging decisions in the general population. Here we tested
Received: 7 April 2023 this pre-registered hypothesis by examining how human participants
Accepted: 12 January 2024 collected resources in an online foraging task. On every trial, participants
chose either to continue to collect rewards from a depleting patch
of resources or to replenish the patch. Participants also completed a
well-validated ADHD self-report screening assessment at the end of
sessions. Participants departed resource patches sooner when travel times
Subject Category:
between patches were shorter than when they were longer, as predicted
Neuroscience and cognition by optimal foraging theory. Participants whose scores on the ADHD
scale crossed the threshold for a positive screen departed patches signifi-
Subject Areas: cantly sooner than participants who did not meet this criterion.
ecology, health and disease and epidemiology, Participants meeting this threshold for ADHD also achieved higher
cognition reward rates than individuals who did not. Our findings suggest that
ADHD attributes may confer foraging advantages in some environments
Keywords: and invite the possibility that this condition may reflect an adaptation
favouring exploration over exploitation.
attention deficit hyperactivity disorder, ADHD,
foraging, behaviour, adaptation,
behavioural biomarker
1. Introduction
All mobile organisms forage for resources such as food, water and mates [1,2].
Authors for correspondence: Foraging for food that is difficult to find or extract has been hypothesized to be
David L. Barack a major driver of the evolution of intelligence [3–5]. Foraging models have
e-mail: dbarack@gmail.com recently been applied to human behaviour in multiple disciplines including
Arjun Ramakrishnan cognitive psychology [6], cognitive neuroscience [7,8] and computer science
[9,10], validating an approach grounded in mathematical ecology to study
e-mail: arjun.ramakrishnan@gmail.com
the computations and mechanisms that evolved to regulate the tradeoff
between exploitation and exploration [11].
Patch foraging is the iterated accept-or-reject decision to stick with a known
† but potentially depleting option or disengage and search for something that
Co-first-authors.
might be better [12]. This description characterizes a wide array of decision
contexts, including when to search for resources like food, water, minerals
Electronic supplementary material is available
or sexual encounters; internal searches through concepts [13], memory [14] or
online at https://doi.org/10.6084/m9.figshare. strategies [15,16]; or searches for abstract resources such as for information
c.7049958. [9,17,18] or reputation. These decisions can be solved using simple algorithms
derived from optimal foraging theory [1], such as the Marginal Value Theorem
Electronic supplementary material is available (MVT) [19]. These algorithms dictate that individuals should leave a depleting
online at https://doi.org/10.6084/m9.figshare. resource patch when local intake rates fall below the average for the environ-
ment. Hundreds of species tested—from bees [20] to birds [21] to monkeys
c.7049958.
[22] to humans [23]—either quantitatively or qualitatively behave in accordance

© 2024 The Author(s) Published by the Royal Society. All rights reserved.
with the predictions of such models, suggesting that evol- nomadic lifestyles that favour exploration [27], endorsing 2
ution long ago settled on a near-optimal solution to the approaches to variation in human behaviour grounded in

royalsocietypublishing.org/journal/rspb
challenge of determining when to abandon a depleting evolutionary medicine [38] and computational psychiatry [39].
resource and search for a new one.
Whether searching for external, internal or abstract
resources, foraging requires balancing a tradeoff between
exploring for new options and exploiting known ones.
2. Methods and materials
Attention deficit hyperactivity disorder (ADHD) is a neurode- (a) Task
velopmental disorder affecting 11% of children [24] and 4.25% We recruited a representative US sample (n = 506) via Prolific
of adults [25]. Symptoms include distractibility as well as rest- (see https://www.prolific.com/) to complete a virtual
lessness and excessive movement [26]. These characteristics patchy foraging task programmed in Javascript. All methods
may impact foraging behaviour by promoting a tendency to were approved by the Internal Review Board of the University
abandon the current resource patch to search for a new one of Pennsylvania. After preprocessing, 49 individuals were
[11]. This tendency to explore while foraging might extend to excluded from final analyses for the following criteria. First,

Proc. R. Soc. B 291: 20222584


other behaviours such as cycling more frequently between they either failed pre-defined survey attention checks or did
information sources in the classroom or sources of stimulation not complete every survey. Further, participants were excluded
in the home environment [11]. Human populations, such as if they completed less than 25 trials (i.e. an exploit or explore
the Ariaal tribe in Africa, that live nomadic lifestyles favouring
Downloaded from https://royalsocietypublishing.org/ on 01 March 2024

decision) throughout the session, not counting incomplete


exploration are characterized by genetic mutations implicated patches (see below). Finally, we excluded participants that
in ADHD [27], consistent with the hypothesis that this con- were believed to have misunderstood the instructions given
dition may impact foraging decisions in other populations. their comments (e.g. ‘I don’t know if I understood it’ or ‘I did
In addition, the neurological and genetic basis of ADHD not understand until the last round’). Note that exclusion cri-
has a well-established noradrenergic component [28–30] teria were slightly changed compared to the pre-registered
linked to task focus [31] and foraging behaviour [32]. Given criteria (see below). Of the final 457 participants (µage =
the link between distractible, impulsive behaviour and atypical 45.63 y.o., σage = 16.09), 232 identified as ‘male’, 217 identified
activity in noradrenergic circuitry, we predict that indivi- as ‘female’, 5 identified as ‘other’, and 3 chose not to provide
duals with ADHD or ADHD-like phenotypes will both leave their gender. The final sample was composed of 71% (n =
patches earlier than predicted by foraging models like 326) White Caucasian (non-Hispanic), 14% (64) African-Amer-
the MVT and also earlier than individuals lacking these ican or Black, 8% (38) Asian, 4% (19) Spanish, Hispanic or
phenotypical characteristics. LatinX, <1% (3) American Indian or Alaskan Native, and less
We sought to test these predictions empirically by study- than 2% (7) unreported.
ing the relationship between foraging behaviour and In the task, a ‘valid patch’ begins with an exploit decision,
ADHD self-report scores in a large population of adults. We in which the user collects berries, and ends with an explore
measured ADHD-like symptoms using the Optimal RiskSLIM decision, in which the user travels to the next patch. Note
DSM-5 ASRS Screening Scale (ASRS) [33]. We designed that any patches where the participant immediately explored
an online incentive-compatible virtual berry-picking game was considered a valid decision, too, and thus included in the
to assess the impact of ADHD symptoms on foraging analysis. Any ‘incomplete patch’ that did not end in the user
behaviour. We tested the pre-registered prediction that stron- leaving the patch (explore) was removed. These incomplete
ger self-reported ADHD symptoms would be associated patches occurred at the end of session blocks in which the
with shorter patch residence times, less optimal behaviour current block timed out after 2 min.
and lower cumulative rewards in our online foraging task. Participants were sent to the training and task via a link in
In our sample of 457 participants, time in patch, travel Qualtrics (see https://www.qualtrics.com/uk/?rid=ip&prev-
time and harvested reward all influenced the decision to site=en&newsite=uk&geo=BE&geomatch=uk) after completing
continue to forage at a patch or to depart and travel to a an informed consent form and several questionnaires. Before
new patch. Overall, participants stayed longer in patches starting, participants received detailed written instructions on
than predicted by MVT. We found that participants with the task and watched a video demonstration. To familiarize
elevated scores on the self-report ADHD scale exhibited themselves with the task, they also completed a 1 min training
shorter patch residence times than those whose scores were (30 sec per block type) before starting the actual experiment.
not elevated. Surprisingly, ADHD screen-positive participants The instructions prior to the training were (emphasis in italics
also achieved higher reward rates and their patch-leaving were coloured in original):
decisions were more closely aligned with those predicted by
the MVT, although they still stayed longer in patches than ‘In this experiment, your goal is to collect as many berries as you
the MVT predicts. can in the amount of time you are given.
ADHD is thus associated with shorter patch dwell times First, hover over the centre box until it fills up completely. You
and higher reward rates in the online foraging environment may have to wiggle your cursor once for the centre box to
implemented here. These findings are consistent with systema- recognize your mouse.
tic amplification of activity within neural circuits involved in Then, on each trial, you will either choose to stay at the
exploration, such as the default mode network (DMN) bush you are at or leave the current bush and travel to a
[12,34], and changes in levels of the neuromodulator norepi- fresh bush.
nephrine broadcast from the locus coeruleus [31,35,36], both To pick berries from the current bush, hover your cursor over
implicated in individuals with ADHD [28–30,37]. Our findings the bush next to the centre box.
resonate with field studies highlighting a potential adaptive fit The number of berries you pick at each attempt will be
between population variation in genes linked to ADHD and indicated next to the bush.
As you keep picking from the same bush repeatedly, the not at liberty to share the method here, but researchers can 3
number of berries you pick at each attempt will go down obtain it after completing a use agreement with the scale

royalsocietypublishing.org/journal/rspb
as you will be depleting this bush. authors (see https://license.tov.med.nyu.edu/product/asrs–
To travel to the next bush, mouse over the box on the opposite dsm-5 for instructions). This is currently free for all academic
side of the bush. purposes. If there are issues obtaining the scoring method,
You can choose to leave the current bush to travel to a fresh researchers could revert to scoring the scale using a simple
bush at any time. When you do that, you will need to wait sum (see electronic supplementary material).
while you travel to the next bush. The ADHD self-report symptom checklist from Ustun et al.
However, once you arrive at the new bush, the amount of ber- [33] used a machine learning algorithm, the risk-calibrated
ries you pick at each harvest will reset to the initial value. super-sparse linear integer model (RiskSLIM), to generate an
You will play different blocks in this game. The environment optimal scoring algorithm for the ADHD Self-Report Screening
will change between these blocks. Scale survey based on the updated DSM-5 criteria for ADHD.
There is an unlimited number of bushes in the environment, The scale was validated by comparison to a managed care
but a finite amount of time. Try to collect as many berries sample of ADHD diagnosed adults (n = 218), a household

Proc. R. Soc. B 291: 20222584


as you can!’ sample from the National Comorbidity Survey Replication
consisting of a face-to-face survey and follow-up (n = 119)
After completing the experiment, participants returned to and an NYU-Langone sample (n = 300). At the recommended
Qualtrics to finish further questionnaires. Foraging data were scoring threshold (≥14) for a positive screen, the ASRS has
Downloaded from https://royalsocietypublishing.org/ on 01 March 2024

stored on a secure AWS server. good sensitivity (91.4%), very good specificity (96.0%), an
During the task, participants collected as many berries as area under curve (AUC) of 0.94, with 67.3% of screen-positive
possible in 8 min (4 blocks, 2 min each). The reward function participants meeting DSM-V criteria for ADHD among a com-
for the task started at 7 plus Gaussian noise, ε, for both long bined cohort from the National Comorbidity Survey-
and short travel time patches for all patches: Replication and a managed care sample, along with good sen-
r0 ¼ 7 þ 1, sitivity (91.9%), moderate specificity (74.0%), an AUC of 0.83
and a proportion of 82.8% of screen-positive participants
where ε ∼ N(0, 0.25) and then, for each choice to harvest who met DSM-V criteria for ADHD among a clinical sample
reward, ri, decremented by 0.5 plus noise: from NYU-Langone Medical Center [33]. The scale has been
ri ¼ ri1  0:5 þ 1, further validated by other groups [40–44].

where again ε ∼ N(0, 0.25). At the end of the experiment, partici- (c) Preregistration
pants received 0.3226 cent per berry collected (up to $3 in total) Our design, hypotheses and analyses were preregistered: see
in addition to their base participation compensation of $4. https://aspredicted.org/blind.php?x=KLK_XLL.
Participants were randomly assigned to one of two orders We made a few justified changes and additions to the pre-
of travel times by block: order 1 [block 1: 1 s travel time, block registered analysis approach. We analysed total reward per
2: 5 s, block 3: 1 s, block 4: 5 s] or order 2 [block 1: 5 s travel patch and reward rates in both conditions (short versus long
time, block 2: 1 s, block 3: 5 s, block 4: 1 s]. Each block lasted travel time) rather than total reward over the entire experiment,
for 2 min. Participants foraged through as many patches as because foraging theory predicts differences between patch
they wished and were able to in those 2 min; patches at the types. We calculated optimal behaviour and deviations from
end of each block that did not terminate with a leave decision optimal behaviour in a way that differed from the preregistered
were removed from analysis. Participants completed an version because we considered the approach by Cowie [45] to
average of 119.95 ± 1.79 trials on average during the task. be more accurate than the pre-registered proposal. Specifically,
The maximum number of trials was 179 and the minimum Cowie’s method presents a first principles way to calculate
was 25 trials, for a range of 154 trials. optimal behaviour. In the binomial regression analysis includ-
ing ASRS, we included ‘cumulative reward in patch’ as
(b) Survey an additional predictor because we considered it relevant to
Participants completed the Optimal RiskSLIM DSM-5 ASRS the model. We did not include trait and state anxiety or positive
Screening Scale [33], consisting of 6 items. The items were: and negative emotions as control variables. The categorization
of participants into ADHD-positive and ADHD-negative
How often do you have difficulty concentrating on what people
are saying to you even when they are speaking to you directly? was not preregistered. However, since the ASRS scale uses a
complex linear scoring algorithm to classify participants as
How often do you leave your seat in meetings or other situations
positive or negative, it made sense to use this qualitative and
in which you are expected to remain seated?
clinically meaningful distinction between participants in our
How often do you have difficulty unwinding and relaxing when analysis. ‘Give up time’ ( pre-registration) is referred to as
you have time to yourself?
‘time in patch’ or ‘patch residence time’ in the current manu-
When you’re in a conversation, how often do you find yourself script, as we considered these to be more accurate terms. The
finishing the sentences of the people you are talking to before
analysis of the link between ASRS with creativity and mind
they can finish them themselves?
wandering will be reported separately for brevity here. Finally,
How often do you put things off until the last minute?
exclusion criteria for participants were slightly improved.
How often do you depend on others to keep your life in order For example, in addition to the pre-registered criteria, we
and attend to details? also excluded participants who commented that they did not
Response options were never, rarely, sometimes, often, and understand the task. Moreover, trials belonging to incomplete
very often. To score the survey, a complex, linear, proprietary patches at the end of each session were excluded before
scoring function was received from its authors [33]. We are applying the ≥25 trials criterion.
(d) Data analysis we used Student’s t-tests to compare ADHD screen-positive 4
All data were analysed in MATLAB (Mathworks, Natick, with screen-negative time in patch, and reward rate

royalsocietypublishing.org/journal/rspb
MA) using custom software. For tests involving comparisons distributions for the two groups were compared using a
across 1 s and 5 s travel times, 1 subject was rejected who Kolmogorov-Smirnov test.
never completed a 5 s travel time patch, leaving n = 456.
For patch residence time comparisons, we averaged the
median patch residence time across subjects, to control for out-
(e) Optimal behaviour
First, an exponential curve was fit to the cumulative intake
liers. We used within-subject paired t-tests to compare the
curve defined by the gain function ( fit function in MATLAB).
average patch residence time for 1 s and 5 s travel time patches.
The gain function g(t) for time in patch t is well-described by
Time in patch was the amount of time from the start of the first
an exponential curve, which can be used to solve from first
trial in a patch to the decision to depart a patch.
principles the optimal leave time [cf. 45]:
For reward-related comparisons, we used both cumulative
reward (which previous work has shown to be the computa- g ¼ b1 (1  eb2 t ): ð2:1Þ
tionally relevant variable for decisions to harvest rewards in

Proc. R. Soc. B 291: 20222584


patch [46]) and reward rate (the classic currency for foraging While this continuous and differentiable function is at
choices in optimal foraging theory [1]). Cumulative reward odds with the discrete real series of rewards, describing the
was defined as the running sum of berries collected within a series of rewards as an exponential permits a first principles
Downloaded from https://royalsocietypublishing.org/ on 01 March 2024

patch, and reward rate was defined by dividing cumulative solution to the optimal patch stay time. To find the optimal
reward by the cumulative time in patch. We used within-subject leave times, the first derivative of g(t) is set equal to the
paired t-tests between travel times to compare cumulative gain function divided by the average travel time and time
rewards and reward rates. We also examined differences in in patch:
cumulative reward across all blocks in the entire session for dg g
ADHD screen-positive and screen-negative participants. ¼ b1 b2 eb2 t ¼ , ð2:2Þ
dt tt þ ^t
To assess the effects of patch number in session, we first
segregated total time in patch and reward rate on patch for travel time tt and optimal leave time ^t. Substituting (2.1) in
leave trials by travel time block (1 s or 5 s). We then separately for g and rearranging terms yields
regressed time in patch and reward rate upon departing ^
eb2 t  b2^t ¼ b2 tt þ 1: ð2:3Þ
patches against patch number in session using ordinary least
squares (OLS) regressions. Patch number in session was We next used the McLaurin expansion for e, eliminated
defined as the patch number in the concatenated blocks constants, dropped the fourth-order and higher terms,
segregated by travel time. subtracted out the first-order term and rearranged to arrive at
ADHD self-report scale (ASRS) scores were correlated
6tt
with both time in patch and reward rates. To determine the 3^t2 þ b2 ^t3  ¼ 0: ð2:4Þ
influence of ASRS on time in patch, we used a mixed-effects b2
linear regression, with random effects of the side of the screen
We used the best-fit β2 from our exponential fit (2.1) to the
of the choice target, patch number in session, and participant
series of average returns. Then, using the average (tt = 3 sec),
identity and fixed effects of travel time, cumulative reward
short (tt = 1 sec) and long (tt = 5 sec) travel times in turn, we
and ASRS score, with all interactions, and fixed effects
determined the optimal travel time ^t by solving (2.4) with
of nuisance covariates age, income and gender, with all
numerical methods ( fmincon in MATLAB subject to nonlinear
interactions. In Wilkinson notation,
constraints determined by the McLaurin expansion).
Time in patch  CumulativeReward  TravelTime
 ASRSScore þ Age  Gender  Income
þ (Side þ PatchNumberinSessionjParticipantNumber), 3. Results
which models patch number in session as a random slope (a) Foraging task and behaviour
and participant identity as a random intercept. To determine Adults (n = 457) performed an online foraging task (figure 1).
the influence of ASRS on decisions to leave patches, we used Participants were recruited via Prolific and were a representa-
a mixed-effects binomial regression, with random effects of tive US sample (47% female, average age = 46y). Participants
the side of the screen of the choice target, patch number were instructed to collect as many berries as possible in 8 min
in session, and participant identity and fixed effects of (4 blocks, 2 min each). Participants first initiated a trial by
time in patch, travel time, cumulative reward and ASRS mousing over an intertrial interval (ITI) box. After initiating
score, with all interactions, and fixed effects of nuisance a trial, participants chose between mousing over an image
covariates age, income and gender, with all interactions. of a bush to stay and collect rewards at a given patch, in
In Wilkinson notation, which payoffs declined with each successive choice, or mous-
ing over an empty-sided box to travel to a new patch, which
~
ProbabilityChooseLeaveTimeInPatch took time. Each participant was tested in two conditions that
 CumulativeReward  TravelTime  ASRSScore differed only in the time it took to ‘travel’ to the next bush
þ Age  Gender  Income (either 1 sec or 5 sec). Travel time was indicated by the
þ (Side þ PatchNumberinSessionjParticipantNumber), visual distance between bushes (see figure 1, inset). The
order of conditions was counterbalanced across participants.
which again models patch number in session as a random After deciding to depart a patch, bushes moved vertically
slope and participant identity as a random intercept. Finally, down the screen, simulating travel, and the participants
stay 5
move cursor to
current bush:

royalsocietypublishing.org/journal/rspb
collect berries
at current patch
+6.75
move cursor to ITI
box to start trial upon cursor contact: upon cursor contact: start new trial in the same
ITI box borders turn collected reward is patch by moving cursor to
dark green after 1 s: displayed for 1 s ITI box
ITI box turns solid green leave
stay or leave?
move cursor
to empty side
travel time to next patch is indicated by distance of box: move to
patches to each other throughout each block a fresh patch
1 s travel time 5 s travel time
2 blocks of 2 min each 2 blocks of 2 min each
upon cursor contact with box: start new trial in the next
bushes move down for 1 s patch by moving cursor to
or 5 s, depending on block, ITI box; current patch is now

Proc. R. Soc. B 291: 20222584


simulating travel on the opposite side
Downloaded from https://royalsocietypublishing.org/ on 01 March 2024

Figure 1. Virtual patch foraging task. Participants began each trial by moving their mouse cursor over an intertrial interval (ITI) box. After 1 s delay, the ITI box turned
green. Participants next decided to stay at a given patch and gather points (’exploit’ decision) or to replenish the patch in exchange for a travel time delay (’explore’
decision). Travel times were either 1 s or 5 s, in blocks. Inset: screenshots of the two foraging environments, with 1 s travel times on the left and 5 s on the right.

(a) 27 (b)
35
26 34

25 33
median total time in patch (s)

rewards collected (arb. units)

32
24
31
23
30
22
29
21
28
20 27
19 26
18 25
1 5 1 5
travel time (s) travel time (s)
Figure 2. Human participants’ (n = 457) behaviour on the virtual patch foraging task. (a) Mean median total time in patch (s) for the 1 s travel time (left) and 5 s
(right). (b) Mean rewards harvested for 1 s travel time (left) and 5 s (right).

faced a new set of decisions between staying in a patch and time environments (RRlong,optimal = 2.2256, p < 1 × 10−119,
departing, with the choice options—the bush and the t (d.f. = 455) = −32.4675). This suboptimal behaviour resulted
empty box switching their sides on the screen. from participants staying for too long in each patch (time
Consistent with predictions from foraging theory’s MVT in patch TiP; short patches: TiPshort = 19.5430 ± 0.1360 s,
[19], participants abandoned patches in a short travel time TiPshort,optimal = 6.0603 s; long patches: TiPlong = 25.5998 ±
(1 sec) environment earlier than they left patches in a long 8039 s, TiPlong,optimal = 12.9049 s), as observed in prior studies
travel time (5 sec) environment (paired t-test on median patch in other animals including nonhuman primates [22,47–50].
residence time by subject, p < 1 × 10−17, t (d.f. = 455) = −9.1942; In addition to these travel time and reward rate effects,
figure 2a). Participants earned significantly more rewards exploratory data analysis uncovered an effect of patch
on average in long travel time environments (paired t-test, p < number in session. Participants performed up to four blocks
1 × 10−44, t (d.f. = 455) = −15.7081; figure 2b), but longer resi- of patches, starting with a block of 1 s patches, then a block
dence times in long travel time patches did not yield of 5 s patches, then 1 s and then 5 s again, or the reverse
significantly higher average reward rates (RR; average order (5, 1, 5, 1; see §2a). Some participants showed an effect
RRshort = 1.4597 ± 0.0271 points s−1, average RRlong = 1.4765 ± of patch number in session on patch residence time and
0.0242 points s−1; paired t-test, p > 0.22, t (d.f. = 455) = −1.2113). reward rate. Segregating by travel time, some subjects
Average reward rates were significantly less than the theor- showed an effect of patch number in session for patch residence
etically optimal reward rates in both short travel time time (1 s patches: OLS, 113/457 participants significant ( p <
environments (RRshort,optimal = 2.6118 points s−1; one-sample 0.05) slope, 74 negative; 5 s patches: OLS, 85/457 participants
t-test, p < 1 × 10−169, t (df = 456) = −45.0179) and long travel significant slope, 53 negative; 168 / 457 unique participants
(a) participant no. 4 participant no. 270 6
30 26
short travel time patch

royalsocietypublishing.org/journal/rspb
long travel time patch 24
total time in patch (s) 25 22

total time in patch (s)


20
20
18
15 16
14
10 12
10
5
8
6

Proc. R. Soc. B 291: 20222584


0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16
patch number in session patch number in session

(b) participant no. 4 participant no. 270


Downloaded from https://royalsocietypublishing.org/ on 01 March 2024

3.0 3.4

reward rate on leaving (berries s–1)


reward rate on leaving (berries s–1)

3.2
2.5
3.0
2.0 2.8
2.6
1.5
2.4
1.0 2.2
2.0
0.5 short travel time patch
long travel time patch 1.8
1.6
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16
patch number in session patch number in session
Figure 3. Participants’ patch residence time and reward rate on leaving a patch as a function of patch number in session, concatenated across blocks for 1 s and 5 s
travel times separately. (a) Total time in patch (s) versus patch number in session, for 1 s (green) and 5 s (red) travel times. Participant no. 4 (left) did not show a
patch number in session effect, whereas participant no. 270 (right) showed a steep decline in total time in patch as a function of patch number in session for both
travel times. (b) Reward rate on patch leave (berries s−1) versus patch number in session for 1 s (green) and 5 s (red) travel times. Participant no. 4 (left) did not
show a change in reward rates as a function of patch number in session, whereas participant no. 270 (right) did, increasing their reward rate.

showed some effect). Two example participants are shown in in session, patch residence time tended to decrease and
figure 3a; the left participant (no. 4) did not show an effect of reward rates increase across sessions; in short, participants
patch number in session (1 s patches: OLS, βslope = 0.1741 ± tended to perform better as they progressed through the task.
0.1490 s/patch number, p > 0.25; 5 s patches: OLS, βslope =
0.1860 ± 0.4653 s/patch number, p > 0.69), whereas the
participant on the right (no. 270) did (1 s patches: OLS, (b) ADHD Self-Report Scale and foraging
βslope = −0.8646 ± 0.1550 s/patch number, p < 0.001; 5 s We next examined how each of the task variables was related
patches: OLS, βslope = −1.5769 ± 0.2684 s/patch number, p < to participants’ ADHD Self-Report Scale scores (ASRS [33]).
0.001). For reward rate, some subjects showed an effect of We first tested for patch-level effects of ASRS by performing
patch number in session (1 s patches: OLS, 131/457 partici- a mixed-effects linear regression of time in patch (n = 10 353
pants significant ( p < 0.05) slope, 126 positive; 5 s patches: patches) against a random effect of patch number in session
OLS, 96/457 participants significant slope, 84 positive; and participant, fixed effects of travel time, cumulative
187/457 unique participants showed some effect). These reward and ASRS scores and all interactions, and a number
same two participants illustrate these effects on reward rate of nuisance covariates including age, gender and income
(figure 3b), with the participant on the left (no. 4) showing and all interactions (see §2d). We reasoned that because of
no effect of patch number in session on reward rate (1 s patches: differences across subjects regarding patch numbers in ses-
OLS, βslope = 0.0114 ± 0.0291 (berries s−1)/patch number, p > sion, covariate should be treated as a random effect. All
0.7; 5 s patches: OLS, βslope = 0.0129 ± 0.0163 (berries s−1)/ fixed effect covariates were z-scored. We uncovered a main
patch number, p > 0.45), unlike the participant on the right effect of cumulative reward ( p ∼ 0, Bonferroni-corrected),
(no. 270) who did (1 s patches: OLS, βslope = 0.0554 ± 0.0113 although there was no main effect of ASRS scores. We
(berries s−1)/patch number, p < 0.001; 5 s patches: OLS, found two-way effects of travel time and cumulative
βslope = 0.1269 ± 0.0174 (berries s−1)/patch number, p < 0.001). reward and of ASRS score and cumulative reward ( p < 5 ×
Of those participants who showed an effect of patch number 10−9 or less, Bonferroni-corrected). Finally, we also found a
regression betas from patch-wise regression against time in patch 7

royalsocietypublishing.org/journal/rspb
travel time

cumulative reward

ASRS score

travel time × cumulative reward

Proc. R. Soc. B 291: 20222584


travel time × ASRS score

cumulative reward × ASRS score


Downloaded from https://royalsocietypublishing.org/ on 01 March 2024

travel time × cumulative reward ×


ASRS score

–0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8


betas
Figure 4. Regression betas for fixed effects of travel time, cumulative reward, and ASRS scores and all interactions on time in patch (n = 10 353 patches). *: p < 0.0005
Bonferroni-corrected.

three-way interaction effect of travel time, cumulative reward participants (Student’s t-test, p < 0.00001, t (df = 455) = 4.5466;
and ASRS score ( p < 0.0005, Bonferroni-corrected). Figure 4 mean reward rate for positive RR+ = 1.5896 ± 0.0356 points s–1;
depicts the regression betas for the covariates of interest. RR– = 1.3608 ± 0.0350), contrary to our prediction. Though
We also examined trial-level effects of ASRS, testing our ASRS screen-positive participants stayed less long in patches
prediction that participants who score highly on the ASRS than screen-negative participants, and earned higher reward
are more likely to choose to leave patches, all else being rates, they still tended to overharvest patches, staying signifi-
equal. We performed a mixed-effects binomial regression of cantly longer than the optimal leave times predicted by the
all decisions to stay (= 0) or leave (= 1) a patch (n = 55 526 MVT for both patch types (one-sample t-test against optimal
trials) against a random effect of patch number in session TiP; mean median TiP for 1 s patches = 17.2336 ± 1.0061 s,
and participant, fixed effects of patch residence time, travel p < 1 × 10−22, t (df = 204) = 11.1059; mean median TiP for 5 s
time, cumulative reward in patch and ASRS, including all patches = 21.2876 ± 0.9211, p < 1 × 10−16, t (df = 204) = 9.0785).
interactions of the fixed effect covariates, and a number Figure 6 illustrates the consistently shorter patch residence
of nuisance covariates and their interactions (see §2d). All times in both conditions for ADHD screen-positive compared
fixed effect covariates were z-scored. Consistent with our pre- to screen-negative participants. Figure 7 illustrates the right-
dictions, all main fixed effects of covariates of interest ( p < shifted distribution of ADHD positive reward rates after
0.005 or less, Bonferroni-corrected) as well as all their two- segregation by positive or negative ASRS screening. Cumulat-
way interactions ( p < 5 × 10−5 or less, Bonferroni-corrected) ive rewards across sessions were significantly greater for
were significant. The three-way interaction of time in patch, ADHD screen-positive than screen-negative participants
cumulative reward and ASRS score was also significant (Student’s t-test, p < 0.0001, t (df = 455) = 4.1000; mean cumulat-
( p < 5 × 10−7, Bonferroni-corrected), as was the four-way ive reward for screen-positive C+ = 602.6460 ± 9.5222;
interaction ( p < 5 × 10−8, Bonferroni-corrected). Figure 5 screen-negative C− = 521.6306 ± 10.0790).
depicts the betas for the covariates of interest. We performed two sets of analyses to provide further
In the light of these significant interaction effects, we next support for our conclusions. First, we ran the same
compared participants who screened positive on the ASRS regressions and approximated the comparisons using OLS
scale (n = 206) with those who screened negative (n = 251), the with a simpler scoring scale for the ASRS screen: the sum
clinically relevant comparison. A positive screen—a score of of the scores of the answers to each item on the questionnaire.
14 or more—indicates that the individual exhibits symptoms The results were almost identical to those reported above (see
consistent with an ADHD diagnosis. Consistent with electronic supplementary material). Second, 24 of our partici-
our predictions, ASRS screen-positive participants stayed in pants reported a previous diagnosis of ADHD (of those 24, 18
patches for significantly shorter durations than screen-negative screened positive using the RiskSLIM scoring sheet [33]). We
(Student’s t-test, p < 0.005, t (d.f. = 455) = −3.0085; mean again ran the same regressions and the analyses as above,
median time in patch for positive TiP+ = 19.3170 ± 0.9057 s; comparing those 24 participants with the 428 participants
TiP– = 23.3907 ± 0.9758 s). However, unexpectedly, they also who reported no previous ADHD diagnosis (5 participants
attained significantly higher reward rates than screen-negative who did not answer that question were left out). The results
regression betas from trial-wise binomial regression against leave decision 8

royalsocietypublishing.org/journal/rspb
time in patch
travel time *
cumulative reward *
ASRS score *
time in patch x travel time *
time in patch × cumulative reward *
travel time × cumulative reward *
time in patch × ASRS score *
travel time × ASRS score *
*

Proc. R. Soc. B 291: 20222584


cumulative reward × ASRS score
time in patch × travel time × *
cumulative reward
time in patch × travel time ×
Downloaded from https://royalsocietypublishing.org/ on 01 March 2024

ASRS score
time in patch × cumulative reward ×
ASRS score
travel time × cumulative reward × *
ASRS score
time in patch × travel time ×
cumulative reward × ASRS score
*
–1.5 –1.0 –0.5 0 0.5 1.0 1.5 2.0 2.5 3.0
betas
Figure 5. Binomial regression betas for fixed effects of time in patch, travel time, cumulative reward and ASRS scores, and all interactions on stay or leave patch
decisions (n = 55 526 trials). *: p < 0.005 Bonferroni-corrected.

median patch residence times by ASRS screen outcomes new one. Participants made these decisions under short or
long travel times between patches, reflecting rich and poor
28
screen – environments. As predicted by the marginal value theorem
screen + (MVT), participants stayed longer in patches when travel
26
median total time in patch (s)

times were longer. Consistent with a range of evidence across


species [45,47,49,51], including humans [52,53], participants
24 stayed longer in patches than predicted by the MVT, potentially
reflecting gathering information about the environment [54], a
22 lack of competing demands due to predation or socializing, or
misrepresentation of rewards [55] or time [49]. The influence of
travel times on patch residence time resulted in significantly
20
different total rewards collected but not significantly different
reward rates between conditions.
18 We next tested for links between foraging decisions and
ADHD Self-Report Scale (ASRS) scores. Correcting for nui-
16 sance covariates, we found that interactions between ASRS
1 5
scores and cumulative reward, travel time and patch number
travel time (s)
in session predicted earlier patch leaving. A trial-level binomial
Figure 6. Patch residence times for ASRS screen-negative (blue) and screen- regression also uncovered significant interactions between
positive (red) participants for 1 s travel time (left) and 5 s travel time (right) ASRS and cumulative reward, patch number in session and
patches. Points are mean median total times in patch, error bars are ±1 s.e.m. time in patch. Comparing ASRS screen-positive with screen-
negative participants revealed shorter patch duration times
and higher reward rates for screen-positive individuals. We
were once again largely confirmatory (see electronic sup- followed-up and supported our findings with two sets of
plementary material), albeit noisier than the outcome of the further analyses, performing the same set of analyses using
first set of confirmatory analyses, perhaps due to so few the sum of the scores of the individual ASRS questions and
participants receiving previous ADHD diagnoses. comparing participants who had received a previous ADHD
diagnosis with those who had not.
There are several strengths and weaknesses to our design.
Our online design enabled gathering ADHD symptom self-
4. Discussion reports from many participants during the height of the
Foraging poses a fundamental challenge for humans and other COVID-19 pandemic. However, our findings relied on self-
organisms [12]. Here, we investigated how humans choose report without monitoring task engagement. Finally, although
between staying in a depleting resource patch or leaving for a the ASRS has been validated [33] and re-validated [40–44] and
distributions of reward rates across all patches 9
0.14

royalsocietypublishing.org/journal/rspb
screen –
screen +
0.12

0.10

frequency 0.08

0.06

Proc. R. Soc. B 291: 20222584


0.04

0.02
Downloaded from https://royalsocietypublishing.org/ on 01 March 2024

0.1

1.1

2.1
0
0.3

1.3

2.3
0.5

1.5

2.5
0.8

1.8

2.8
0.7

1.7

2.7
0.9

1.9

2.9
0.6

1.6

2.6
0.2

1.2

2.2
0.4

1.4

2.4
1.0

2.0

3.0
observed reward rate (pts s–1)
Figure 7. Distribution of patch reward rates across all patches for ASRS screen-negative (blue) and screen-positive (red) participants. Large peaks at 0 are from
participants departing patches immediately upon entry. Vertical dashed lines are optimal reward rates (see §2e): red: 1 s travel time optimal reward rate; black:
global optimal reward rate; green: 5 s travel time optimal reward rate.

self-report scales have been found to be informative for detect- and continued persistence in the human population. Individ-
ing ADHD [56–58], clinical validation is still needed to confirm uals with ADHD may be more reward-seeking, consistent
the relationship between foraging behaviour and ADHD. with the role of dopamine in motivated behaviour [72] as
In our sample, 206 of 457 participants, or 45%, screened well as heightened impulsivity on delayed discounting tasks
positive for ADHD, approximately 10x higher than the rate [73]. Individuals with ADHD may also be more exploratory
of ADHD in the adult population. Several factors may account as a result of increased noradrenergic drive originating in the
for this discrepancy. First, ADHD-screen positive participants locus coeruleus and broadcast to prefrontal cortex [29]. Typical
were not diagnosed with ADHD; only a clinical assessment individuals, by contrast, may be driven by a variety of motiv-
can confirm such a diagnosis. Second, we conducted our ations, including gathering more information about the
study online using Prolific, which may have elevated rates of environment through persistent foraging.
positive screens, as was found using an older ADHD screen Our task focuses on core explore–exploit foraging
in a previous mTurk sample [59]. Finally, our data collection decisions, a type of decision ubiquitous in the real world. For
occurred during the COVID-19 pandemic, which negatively example, the Aché people of Paraguay hunt in accordance
impacted mental health for broad swaths of the public [60], with the predictions of optimal foraging theory [74], as do
as confirmed by online self-reports [61]. Notably, the pandemic Nahua mushroom foragers [75], and human shopping behav-
is linked to a global increase in ADHD symptoms [62]. While iour is also consistent with optimal foraging predictions [76–
difficult to estimate, this context undoubtedly played a role 78]. However, there are key differences between naturalistic
in the frequency of ADHD symptoms reported in our sample. foraging behaviour and patch leaving tasks like ours. Notably,
Higher reward rates in ADHD screen-positive participants there are minor metabolic costs to our task, whereas naturalistic
may reflect adaptive specialization in the capacity to focus. foraging involves substantial metabolic costs associated with
Attention is known to be regulated in part by dopaminergic leaving a patch and searching for a new one. In addition,
receptors and transporters [63,64]. Animal models of ADHD our task required participants to gather points—an abstract
show dopaminergic deficits in the basal ganglia and prefrontal resource that stands in contrast to basic alimentary or
cortex [65], and both the D2 [66] and the D4 [67] dopamine other resources like calories. Nonetheless, animals and
receptors are associated with ADHD. These same receptors humans show similar patterns of behaviour that accord with
are implicated in differences in nutrition and health in compari- foraging theory, whether foraging for resources in the real
sons of nomadic and settled populations [27]; specifically, the world or foraging for food pellets or points on a computer in
interaction between lifestyle and allele length for the D4 recep- the laboratory [22,50,55,79].
tor predicts healthier body mass. Longer alleles for the D4 Our task was designed to understand how participants
receptor are a risk factor for ADHD [67]. Taken together, make patch leaving decisions and whether participants with
allele length for the D4 receptor is a risk factor for ADHD different cognitive phenotypes exhibit variability in those
and correlates with better nutritive outcomes in nomadic decisions. These decisions are characterized by tracking
populations. Consistent with our reported findings, we specu- resource intakes and deciding to replenish that resource. In
late that ADHD serves as an adaptive specialization for the real world, this decision would involve searching for a
foraging [68–71], thus explaining its widespread prevalence new resource instead of merely resetting it. Unlike the real
world, our task does not require search (cf. the distinction prey types places attentional demands that lead to selection of 10
between searching and harvesting in [80]). Nonetheless, by prey of a particular type [87], consistent with diet selection

royalsocietypublishing.org/journal/rspb
focusing on patch leaving decisions, our task provides funda- rules in optimal foraging theory [1]. In addition, a comparison
mental insights on foraging choices without the confounding of feature search and conjunctive search revealed that search
factors involved in searching for new patches. type augments attentional demands and consequently fora-
ADHD-like cognitive phenotypes confer advantages or ging costs, resulting in differences in give-up times [88], a key
disadvantages depending on the environment. Foragers with foraging measure [89]. Finally, search paths are longer and
ADHD-like traits may fare poorly in environments with mul- more variable in individuals with greater ADHD symptoms
tiple, learnable depletion rates (like the task used in [79]), [90]. Our findings build on this literature, suggesting
by leaving patches before gathering enough information to that heightened ADHD traits may confer certain foraging
learn the different patch types. In environments in which advantages by lowering exploration thresholds.
exhausted patches are renewed, primates are known to track The DMN, a distributed neural circuit with high resting-
renewal rates, such as the availability of fruit [81–83]. In these state activity [28,29], shows elevated activity during task per-
environments, ADHD-type foragers may be likelier to visit formance in individuals with ADHD [30,31], potentially

Proc. R. Soc. B 291: 20222584


resources too early or too late due to noisier estimates of accounting for more frequent errors [29]. Neurons in posterior
depletion rates, resulting in fewer rewards. However, in cingulate cortex (PCC), a central node of the DMN, signal
competitive environments where foragers must keep track of decisions to explore [33] and the information driving these
other foragers, impulsively leaving patches could yield a decisions [18]. In addition, electrical microstimulation in PCC
Downloaded from https://royalsocietypublishing.org/ on 01 March 2024

competitive advantage by enabling learning about competitors promotes exploration of a non-default option [34]. ADHD is
and capturing newly renewed resources first. associated with noradrenergic activity [28–30], and stimulant
Many decision contexts differ from foraging ones by pre- medications like reboxetine [91] or atomoxetine [92] target
senting agents with multiple options simultaneously rather these systems. Notably, stimulation of the locus coeruleus—
than single depleting resources. We hypothesize that individ- the source of noradrenaline in the brain—provokes early
uals with ADHD-type traits may struggle more in these patch leaving in rats [32], and treatment with reboxetine in
contexts, a proposal backed by existing research [84]. For humans induces earlier patch departures, higher reward rates
instance, a study by Frank et al. [85] used a reinforcement learn- at patch leaves and more optimal foraging [50]. The locus coer-
ing task to assess learning in neurotypical (healthy) individuals, uleus projects to a number of DMN nodes, including posterior
ADHD patients on medication and ADHD patients off medi- cingulate cortex [93,94]. Together, these findings suggest that
cation. The participants had to learn which of two arbitrary noradrenergic activity ‘tunes’ circuits regulating explore–
symbols was associated with more rewards by learning the exploit decisions, resulting in advantages for ADHD-type indi-
expected value of selecting each option. In their task, outcomes viduals in foraging.
were independent of previous choices, there was no depleting In conclusion, our virtual foraging task reveals that people
resource and choices were between options instead of accepting adaptively follow predictions of optimal foraging theory. They
or rejecting a single offer. They found that patients off medi- adapted their foraging strategies to the local statistics of patches
cation displayed lower choice accuracy and switched choices and stayed longer in long travel time patches compared to
more frequently after rewarding feedback than patients on shorter travel times, as predicted by optimal foraging theory.
medication or healthy controls. This study illustrates how Nevertheless, people generally overharvested patches and con-
in an environment with symbolic associations between rewards sequently did not conform to the optimal patch-leaving time
and options, switching leads to maladaptive information specified by foraging theory. In addition, we discovered that
gathering to learn those associations, in contrast to a patchy participants that screened positive for ADHD more readily
foraging environment like our task, where switching behaviour abandoned patches and achieved higher reward rates than
may be adaptive. did participants who screened negative. Given the over-staying
As a second example, a study by Dekkers et al. [86] investi- displayed by participants overall, those with elevated ASRS
gated how ADHD participants and healthy controls made scores made exploratory decisions that were more closely
choices between a more certain ‘safe’ option and a ‘risky’ aligned with the predictions of optimal foraging theory, and,
option, each with explicitly verbally communicated expected in this sense, behaved more optimally. The increased foraging
values. On some trials the safe option had the higher expected proficiency of participants with ADHD-like behaviour
value, whereas on others the risky option had a higher observed here suggests that the prevalence and persistence of
expected value. ADHD participants less frequently selected ADHD in human populations may serve an adaptive function
the option with the higher expected value. This decision con- in some environments.
text involves a two-alternative forced choice where each
decision was independent and non-depleting, differing mark- Ethics. All methods were approved by the Internal Review Board of
the University of Pennsylvania (IRB #823436).
edly from the patch foraging task. Here, neither switching
Data accessibility. All data are available at https://doi.org/10.5281/
nor staying was an intrinsically adaptive strategy, and optimal zenodo.10178228 [95].
decisions required information gathering and deployment (i.e. Supplementary material is available online [96].
in reading and computing the expected value of options). Declaration of AI use. We have not used AI-assisted technologies in
Modern decision contexts often involve non-depleting, inde- creating this article.
pendent choices, usually framed in symbolic terms. Such Authors’ contributions. D.L.B.: data curation, formal analysis, software,
tasks may pose specific challenges for individuals with writing—original draft, writing—review and editing; V.U.L.: concep-
ADHD-type traits, whose decision-making mechanisms may tualization, investigation, methodology, project administration,
supervision, writing—original draft, writing—review and editing;
be better suited for foraging-like environments. F.P.: data curation, writing—review and editing; N.A.: investigation,
A burgeoning literature reveals links between foraging pro- supervision; E.M.B.: writing—original draft, writing—review and
ficiency and attention to ADHD-like traits. Foraging for cryptic editing; A.R.: conceptualization, software; M.L.P.: conceptualization,
funding acquisition, project administration, writing—original draft, Acknowledgements. The Wharton Behavioural Lab provided support 11
writing—review and editing. in setting up and running the study. We are grateful to Emily
All authors gave final approval for publication and agreed to be M. Orengo and Richard Lee for their assistance during this project.

royalsocietypublishing.org/journal/rspb
held accountable for the work performed therein. The research was supported by R37-MH109728, R01-MH108627,
Conflict of interest declaration. We declare we have no competing interests. R01-MH-118203, KA2019-105548, U01MH121260, UM1MH130981,
Funding. The authors report no financial conflicts of interest or R56MH122819, R56AG071023 (all to M.L.P.), the Wharton Behavioural
disclosures. Lab and the Wharton Dean’s Research Fund (to V.U.L. and M.L.P.).

References
1. Stephens DW, Krebs JR. 1986 Foraging theory. Topics Cogn. Sci. 7, 513–534. (doi:10.1111/tops. among nomadic and recently settled Ariaal men of
Princeton, NJ: Princeton University Press. 12151) northern Kenya. BMC Evol. Biol. 8, 1–12. (doi:10.
2. Stephens DW et al. 2007 Foraging: behavior and 15. Metcalfe J, Jacobs WJ. 2010 People’s study time 1186/1471-2148-8-173)

Proc. R. Soc. B 291: 20222584


ecology. Chicago, IL: University of Chicago Press. allocation and its relation to animal foraging. 28. Biederman J, Spencer T. 1999 Attention-deficit/
3. Milton K. 1988 Foraging behaviour and the Behav. Processes 83, 213–221. (doi:10.1016/j. hyperactivity disorder (ADHD) as a noradrenergic
evolution of primate intelligence. In Machiavellian beproc.2009.12.011) disorder. Biol. Psychiatry 46, 1234–1242. (doi:10.
intelligence: social expertise and the evolution of 16. Payne S, Duggan G. 2011 Giving up problem 1016/S0006-3223(99)00192-4)
Downloaded from https://royalsocietypublishing.org/ on 01 March 2024

intellect in monkeys, apes, and humans (ed. AWRW solving. Mem. Cognit. 39, 902–913. (doi:10.3758/ 29. Arnsten AF. 2000 Genetics of childhood disorders:
Byrne), pp. 285–305. New York, NY: Clarendon s13421-010-0068-6) XVIII. ADHD, Part 2: Norepinephrine has a
Press/Oxford University Press. 17. Pirolli P, Card S. 1999 Information foraging. critical modulatory influence on prefrontal cortical
4. Genovesio A, Wise SP, Passingham RE. 2014 Psychol. Rev. 106, 643. (doi:10.1037/0033-295X. function. J. Am. Acad. Child Adolesc. Psychiatry
Prefrontal–parietal function: from foraging to 106.4.643) 39, 1201–1203. (doi:10.1097/00004583-
foresight. Trends Cogn. Sci. 18, 72–81. (doi:10.1016/ 18. Barack DL, Bakkour A, Shohamy D, Salzman CD. 200009000-00022)
j.tics.2013.11.007) 2021 Signatures of visuospatial information foraging 30. Xu X et al. 2008 Replication of a rare protective
5. DeCasien AR, Williams SA, Higham JP. 2017 during learning of complex environments. bioRxiv. allele in the noradrenaline transporter gene and
Primate brain size is predicted by diet but not (doi:10.1101/2021.09.22.461356) ADHD. Am. J. Med. Genet. B: Neuropsych. Genet.
sociality. Nat. Ecol. Evol., 1, 0112. (doi:10.1038/ 19. Charnov EL. 1976 Optimal foraging, the marginal 147, 1564–1567. (doi:10.1002/ajmg.b.30872)
s41559-017-0112) value theorem. Theor. Popul. Biol. 9, 129–136. 31. Aston-Jones G, Cohen JD. 2005 An integrative
6. Todd PM, Hills TT. 2020 Foraging in mind. Curr. Dir. (doi:10.1016/0040-5809(76)90040-X) theory of locus coeruleus-norepinephrine function:
Psychol. Sci. 29, 309–315. (doi:10.1177/ 20. Waddington KD, Holden LR. 1979 Optimal foraging: adaptive gain and optimal performance. Annu. Rev.
0963721420915861) on flower selection by bees. Am. Nat. 114, Neurosci. 28, 403–450. (doi:10.1146/annurev.neuro.
7. Calhoun AJ, Hayden BY. 2015 The foraging brain. 179–196. (doi:10.1086/283467) 28.061604.135709)
Curr. Opin. Behav. Sci. 5, 24–31. (doi:10.1016/j. 21. Krebs JR, Kacelnik A, Taylor P. 1978 Test of optimal 32. Kane GA, Vazey EM, Wilson RC, Shenhav A, Daw ND,
cobeha.2015.07.003) sampling by foraging great tits. Nature 275, 27–31. Aston-Jones G, Cohen JD. 2017 Increased locus
8. Mobbs D, Trimmer PC, Blumstein DT, Dayan P. 2018 (doi:10.1038/275027a0) coeruleus tonic activity causes disengagement from
Foraging for foundations in decision neuroscience: 22. Hayden BY, Pearson JM, Platt ML. 2011 Neuronal a patch-foraging task. Cogn. Affect. Behav. Neurosci.
insights from ethology. Nat. Rev. Neurosci. 19, basis of sequential foraging decisions in a patchy 17, 1073–1083. (doi:10.3758/s13415-017-0531-y)
419–427. (doi:10.1038/s41583-018-0010-7) environment. Nat. Neurosci. 14, 933–939. (doi:10. 33. Ustun B, Adler LA, Rudin C, Faraone SV, Spencer TJ,
9. Pirolli PLT. 2007 Information foraging theory: 1038/nn.2856) Berglund P, Gruber MJ, Kessler RC. 2017 The World
adaptive interaction with information. New York, NY: 23. Hawkes K, Hill K, O’Connell JF. 1982 Why hunters Health Organization adult attention-deficit/
Oxford University Press. gather: optimal foraging and the Ache of eastern hyperactivity disorder self-report screening scale for
10. Fu W-T, Pirolli P. 2007 SNIF-ACT: A cognitive model Paraguay. Am. Ethnol. 9, 379–398. (doi:10.1525/ae. DSM-5. Jama Psychiatry 74, 520–526. (doi:10.1001/
of user navigation on the World Wide Web. Hum.– 1982.9.2.02a00100) jamapsychiatry.2017.0298)
Comp. Interact. 22, 355–412. (https://psycnet.apa. 24. Visser SN, Danielson ML, Bitsko RH, Holbrook JR, 34. Raichle ME. 2015 The brain’s default mode network.
org/record/2007-18250-001) Kogan MD, Ghandour RM, Perou R, Blumberg SJ. Annu. Rev. Neurosci. 38, 433–447. (doi:10.1146/
11. Addicott M, Pearson JM, Sweitzer MM, Barack DL, 2014 Trends in the parent-report of health care annurev-neuro-071013-014030)
Platt M. 2017 A primer on foraging and the provider-diagnosed and medicated attention-deficit/ 35. Nassar MR, Rumsey KM, Wilson RC, Parikh K,
explore/exploit trade-off for psychiatry research. hyperactivity disorder: United States, 2003–2011. Heasly B, Gold JI. 2012 Rational regulation of
Neuropsychopharmacology 42, 1931–1939. (doi:10. J. Am. Acad. Child Adolesc. Psychiatry 53, 34–46.e2. learning dynamics by pupil-linked arousal
1038/npp.2017.108) (doi:10.1016/j.jaac.2013.09.001) systems. Nat. Neurosci. 15, 1040–1046. (doi:10.
12. Barack DL, Platt ML. 2017 Engaging and Exploring: 25. London AS, Landes SD. 2021 Cohort Change in the 1038/nn.3130)
Cortical Circuits for Adaptive Foraging Decisions. In Prevalence of ADHD Among U.S. Adults: Evidence of 36. Warren CM, Wilson RC, Van Der Wee NJ, Giltay EJ,
Impulsivity (ed. J Stevens), pp. 163–199. a Gender-Specific Historical Period Effect. J. Atten. Van Noorden MS, Cohen JD, Nieuwenhuis S. 2017
Berlin, Germany: Springer. (doi:10.1007/978-3-319- Disord. 25, 771–782. (doi:10.1177/ The effect of atomoxetine on random and directed
51721-6_6) 1087054719855689) exploration in humans. PloS ONE 12, e0176034.
13. Hills TT, Todd PM, Goldstone RL. 2008 Search in 26. Posner J, Polanczyk GV, Sonuga-Barke E. 2020 (doi:10.1371/journal.pone.0176034 )
external and internal spaces evidence for Attention-deficit hyperactivity disorder. Lancet 395, 37. Metin B, Krebs RM, Wiersema JR, Verguts T,
generalized cognitive search processes. Psychol. 450–462. (doi:10.1016/S0140-6736(19)33004-1) Gasthuys R, van der Meere JJ, Achten E, Roeyers H,
Sci. 19, 802–808. (doi:10.1111/j.1467-9280. 27. Eisenberg DT, Campbell B, Gray PB, Sorenson MD. Sonuga-Barke E. 2015 Dysfunctional modulation of
2008.02160.x) 2008 Dopamine receptor genetic polymorphisms default mode network activity in attention-deficit/
14. Hills TT, Todd PM, Jones MN. 2015 Foraging in and body composition in undernourished hyperactivity disorder. J. Abnorm. Psychol. 124, 208.
semantic fields: How we search through memory. pastoralists: An exploration of nutrition indices (doi:10.1037/abn0000013)
38. Stearns SC. 2012 Evolutionary medicine: its scope, 51. Hubbard S, Cook R. 1978 Optimal foraging by 65. Russell V, de Villiers A, Sagvolden T, Lamm M, 12
interest and potential. Proc. R. Soc. B 279, parasitoid wasps. J. Anim. Ecol. 47, 593–604. Taljaard J. 1995 Altered dopaminergic function in

royalsocietypublishing.org/journal/rspb
4305–4321. (doi:10.1098/rspb.2012.1326) (doi:10.2307/3803) the prefrontal cortex, nucleus accumbens and
39. Huys QJ, Maia TV, Frank MJ. 2016 Computational 52. Hutchinson JM, Wilke A, Todd PM. 2008 Patch caudate-putamen of an animal model of
psychiatry as a bridge from neuroscience to clinical leaving in humans: can a generalist adapt its attention-deficit hyperactivity disorder—the
applications. Nat. Neurosci. 19, 404–413. (doi:10. rules to dispersal of items across patches? Anim. spontaneously hypertensive rat. Brain Res.
1038/nn.4238) Behav. 75, 1331–1349. (doi:10.1016/j.anbehav. 676, 343–351. (doi:10.1016/0006-8993(95)
40. Silverstein MJ, Alperin S, Faraone SV, Kessler RC, 2007.09.006) 00135-D)
Adler LA. 2018 Test–retest reliability of the adult 53. Constantino SM, Daw ND. 2015 Learning the 66. Moro MM, El-Gebaly HH, Zaky EA, Kamal TM. 2019
ADHD Self-Report Scale (ASRS) v1.1 Screener in opportunity cost of time in a patch-foraging task. A study of dopamine D2 receptor Taq1 a alleles in
non-ADHD controls from a primary care physician Cogn. Affect. Behav. Neurosci. 15, 837–853. (doi:10. children with attention-deficit hyperactivity
practice. Fam. Pract. 35, 336–341. (doi:10.1093/ 3758/s13415-015-0350-y) disorder. Curr. Pediatr. Res 23, 9–16.
fampra/cmx115) 54. Davidson JD, El Hady A. 2019 Foraging as an 67. Bonvicini C, Cortese S, Maj C, Baune BT, Faraone SV,
41. Green JG, DeYoung G, Wogan ME, Wolf EJ, Lane KL, evidence accumulation process. PLoS Comput. Biol. Scassellati C. 2020 DRD4 48 bp multiallelic variants

Proc. R. Soc. B 291: 20222584


Adler LA. 2019 Evidence for the reliability and 15, e1007060. (doi:10.1371/journal.pcbi.1007060) as age-population-specific biomarkers in attention-
preliminary validity of the Adult ADHD Self-Report 55. Kane GA, Bornstein AM, Shenhav A, Wilson RC, Daw deficit/hyperactivity disorder. Translational
Scale v1.1 (ASRS v1.1) Screener in an adolescent ND, Cohen JD. 2019 Rats exhibit similar biases in Psychiatry 10, 1–19. (doi:10.1038/s41398-020-
community sample. Int. J. Methods Psychiatr. Res. foraging and intertemporal choice tasks. Elife 8, 0755-4)
Downloaded from https://royalsocietypublishing.org/ on 01 March 2024

28, e1751. (doi:10.1002/mpr.1751) e48429. (doi:10.7554/eLife.48429) 68. Rozin P, Schull J et al. 1988 The adaptive-
42. Silverstein MJ, Faraone SV, Alperin S, Leon TL, 56. Abrams J, Faraone SV, Woodworth KY, Spencer TJ, evolutionary point of view in experimental
Biederman J, Spencer TJ, Adler LA. 2019 Validation Biederman I, Biederman J. 2018 Are Adult ADHD psychology. In Stevens’ handbook of experimental
of the Expanded Versions of the Adult ADHD Self- Patients Good Informants of Their Symptoms? A psychology: perception and motivation; learning and
Report Scale v1.1 Symptom Checklist and the Adult Qualitative Literature Review of Concordance cognition (ed. RC Atkinson), pp. 503–546.
ADHD Investigator Symptom Rating Scale. J. Atten. Between Clinician and Self-Report ADHD Symptoms. New York, NY: John Wiley & Sons.
Disord. 23, 1101–1110. (doi:10.1177/ J. Nerv. Ment. Dis. 206, 739–743. (doi:10.1097/ 69. Krebs JR. 1990 Food-storing birds: adaptive
1087054718756198) NMD.0000000000000870) specialization in brain and behaviour? Phil. Trans. R.
43. Brevik EJ, Lundervold AJ, Haavik J, Posserud MB. 57. Silverstein MJ, Faraone SV, Alperin S, Biederman J, Soc. B 329, 153–160. (doi:10.1098/rstb.1990.0160)
2020 Validity and accuracy of the adult attention- Spencer TJ, Adler LA. 2018 How informative are 70. Kacelnik A. 2006 Meanings of rationality. In Rational
deficit/hyperactivity disorder (ADHD) self-report self-reported adult attention-deficit/hyperactivity Animals (eds S Hurley, M Nudds), pp. 87–106.
scale (ASRS) and the Wender Utah rating scale disorder symptoms? An examination of the Oxford, UK: Oxford Academic. (doi:10.1093/acprof:
(WURS) symptom checklists in discriminating agreement between the adult attention-deficit/ oso/9780198528272.003.0002)
between adults with and without ADHD. Brain hyperactivity disorder self-report scale V1. 1 and 71. Jensen PS, Mrazek D, Knapp PK, Steinberg L, Pfeffer
Behav. 10, e01605. (doi:10.1002/brb3.1605) adult attention-deficit/hyperactivity disorder C, Schowalter J, Shapiro T. 1997 Evolution and
44. Somma A, Adler LA, Gialdi G, Arteconi M, Cotilli E, investigator symptom rating scale. J. Child Adolesc. revolution in child psychiatry: ADHD as a disorder of
Fossati A. 2021 The validity of the World Health Psychopharmacol. 28, 339–349. (doi:10.1089/cap. adaptation. J. Am. Acad. Child Adolesc. Psych. 36,
Organization Adult attention-deficit/hyperactivity 2017.0082) 1672–1681. (doi:10.1097/00004583-199712000-
disorder self-report screening scale for diagnostic 58. Anbarasan D, Kitchin M, Adler LA. 2020 Screening 00015)
and statistical manual of mental disorders, in for Adult ADHD. Curr. Psychiatry Rep. 22, 72. 72. Schultz W. 2022 Dopamine reward prediction error
adolescence. J. Child Adolesc. Psychopharmacol. 31, (doi:10.1007/s11920-020-01194-9) coding. Dialogues Clin. Neurosci. 18, 23–32. (doi:10.
631–638. (doi:10.1089/cap.2020.0158) 59. Wymbs BT, Dawson AE. 2019 Screening Amazon’s 31887/dcns.2016.18.1/wschultz)
45. Cowie RJ. 1977 Optimal foraging in great tits (Parus Mechanical Turk for Adults With ADHD. J. Atten. 73. Marx I, Hacker T, Yu X, Cortese S, Sonuga-Barke E.
major). Nature 268, 137–139. (doi:10.1038/ Disord. 23, 1178–1187. (doi:10.1177/ 2021 ADHD and the choice of small immediate over
268137a0) 1087054715597471) larger delayed rewards: a comparative meta-analysis
46. Barack DL, Chang SW, Platt ML. 2017 Posterior 60. Usher K, Durkin J, Bhullar N. 2020 The COVID-19 of performance on simple choice-delay and
cingulate neurons dynamically signal decisions to pandemic and mental health impacts. Int. J. Ment. temporal discounting paradigms. J. Atten. Disord.
disengage during foraging. Neuron 96, 339–347.e5. Health Nurs. 29, 315. (doi:10.1111/inm.12726) 25, 171–187. (doi:10.1177/1087054718772138)
(doi:10.1016/j.neuron.2017.09.048) 61. Jewell JS, Farewell CV, Welton-Mitchell C, Lee-Winn 74. Hill K, Kaplan H, Hawkes K, Hurtado AM. 1987
47. Pyke GH, Pulliam HR, Charnov EL. 1977 Optimal A, Walls J, Leiferman JA. 2020 Mental health during Foraging decisions among Ache hunter-gatherers:
foraging: a selective review of theory and tests. Q. the COVID-19 pandemic in the United States: online new data and implications for optimal foraging
Rev. Biol. 52, 137–154. (doi:10.1086/409852) survey. JMIR Formative Research 4, e22043. (doi:10. models. Ethol. Sociobiol. 8, 1–36. (doi:10.1016/
48. Nonacs P. 2001 State dependent behavior and the 2196/22043) 0162-3095(87)90055-0)
marginal value theorem. Behav. Ecol. 12, 71–83. 62. Rogers MA, MacLean J. 2023 ADHD Symptoms 75. Pacheco-Cobos L, Winterhalder B, Cuatianquiz-Lima
(doi:10.1093/oxfordjournals.beheco.a000381) Increased During the Covid-19 Pandemic: A Meta- C, Rosetti MF, Hudson R, Ross CT. 2019 Nahua
49. Kendall RK, Wikenheiser AM. 2022 Quitting while Analysis. J. Atten. Disord. 27, 800–811. (doi:10. mushroom gatherers use area-restricted search
you’re ahead: Patch foraging and temporal 1177/10870547231158750) strategies that conform to marginal value theorem
cognition. Behav. Neurosci. 136, 467–478. (doi:10. 63. Swanson JM et al. 2000 Dopamine genes and predictions. Proc. Natl Acad. Sci. USA 116, 10
1037/bne0000526) ADHD. Neuroscience & Biobehavioral Reviews 24, 339–10 347. (doi:10.1073/pnas.1814476116)
50. Sidorenko N, Chung HK, Grueschow M, Quednow 21–25. (doi:10.1016/S0149-7634(99)00062-7) 76. Rajala AK, Hantula DA. 2000 Towards a behavioral
BB, Hayward-Könnecke H, Jetter A, Tobler PN. 64. Li D, Sham PC, Owen MJ, He L. 2006 Meta-analysis ecology of consumption: Delay-reduction effects on
2023 Acetylcholine and noradrenaline enhance shows significant association between dopamine foraging in a simulated Internet mall. Manage.
foraging optimality in humans. Proc. Natl Acad. Sci. system genes and attention deficit hyperactivity Decis. Econ. 21, 145–158. (doi:10.1002/mde.979)
USA 120, e2305596120. (doi:10.1073/pnas. disorder (ADHD). Hum. Mol. Genet. 15, 2276–2284. 77. Foxall GR, James VK. 2003 The behavioral ecology
2305596120 ) (doi:10.1093/hmg/ddl152) of brand choice: How and what do consumers
maximize? Psychol. Market. 20, 811–836. (doi:10. 84. Mowinckel AM, Pedersen ML, Eilertsen E, Biele G. increased exploratory foraging patterns. Sci. Rep. 9, 13
1002/mar.10098) 2015 A meta-analysis of decision-making 1–7. (doi:10.1038/s41598-019-46761-0)

royalsocietypublishing.org/journal/rspb
78. Hantula DA, Brockman DD, Smith CL. 2008 Online and attention in adults with ADHD. J. Atten. 91. Ghanizadeh A. 2015 A systematic review of
shopping as foraging: The effects of increasing Disord. 19, 355–367. (doi:10.1177/10870547 reboxetine for treating patients with attention
delays on purchasing and patch residence. IEEE 14558872) deficit hyperactivity disorder. Nord. J. Psychiatry 69,
Trans. Prof. Commun. 51, 147–154. (doi:10.1109/ 85. Frank MJ, Santamaria A, O’Reilly RC, Willcutt E. 241–248. (doi:10.3109/08039488.2014.972975)
TPC.2008.2000340) 2007 Testing computational models of dopamine 92. Michelson D et al. 2003 Atomoxetine in adults with
79. Harhen NC, Bornstein AM. 2023 Overharvesting in and noradrenaline dysfunction in attention deficit/ ADHD: two randomized, placebo-controlled studies.
human patch foraging reflects rational structure hyperactivity disorder. Neuropsychopharmacology Biol. Psychiatry 53, 112–120. (doi:10.1016/S0006-
learning and adaptive planning. Proc. Natl Acad. Sci. 32, 1583–1599. (doi:10.1038/sj.npp.1301278) 3223(02)01671-2)
USA 120, e2216524120. (doi:10.1073/pnas. 86. Dekkers TJ, Agelink van Rentergem JA, Huizenga 93. Levitt P, Moore RY. 1978 Noradrenaline neuron
2216524120) HM, Raber H, Shoham R, Popma A, Pollak Y. 2021 innervation of the neocortex in the rat. Brain
80. Hills TT, Todd PM, Goldstone RL. 2010 The central Decision-making deficits in ADHD are not related to Res. 139, 219–231. (doi:10.1016/0006-
executive as a search process: priming exploration risk seeking but to suboptimal decision-making: 8993(78)90925-3)

Proc. R. Soc. B 291: 20222584


and exploitation across domains. J. Exp. Psychol. Meta-analytical and novel experimental evidence. 94. Heilbronner SR, Haber SN. 2014 Frontal cortical and
Gen. 139, 590. (doi:10.1037/a0020666) J. Atten. Disord. 25, 486–501. (doi:10.1177/ subcortical projections provide a basis for
81. Terborgh J. 1983 Five New World primates: A study 1087054718815572) segmenting the cingulum bundle: implications for
in comparative ecology. Monographs in Behavior and 87. Kristjánsson Á, Jóhannesson ÓI, Thornton IM. 2014 neuroimaging and psychiatric disorders. J. Neurosci.
Downloaded from https://royalsocietypublishing.org/ on 01 March 2024

Ecology, Vol. 46. Princeton, NJ: Princeton University Common attentional constraints in visual foraging. 34, 10 041–10 054. (doi:10.1523/jneurosci.5459-
Press. (doi:10.2307/j.ctt7ztr6n) PloS ONE 9, e100752. (doi:10.1371/journal.pone. 13.2014)
82. Watts DP. 1998 Long-term habitat use by mountain 0100752) 95. Barack DL, Ludwig VU, Parodi F, Ahmed N,
gorillas (Gorilla gorilla beringei). 2. Reuse of 88. Kristjánsson T, Thornton IM, Chetverikov A, Brannon EM, Ramakrishnan A, Platt ML. 2024
foraging areas in relation to resource abundance, Kristjánsson Á. 2020 Dynamics of visual attention Attention deficits linked with proclivity to explore
quality, and depletion. Int. J. Primatol. 19, revealed in foraging tasks. Cognition 194, 104032. while foraging. Zenodo. (doi:10.5281/zenodo.
681–702. (doi:10.1023/A:1020376925939) (doi:10.1016/j.cognition.2019.104032) 10178228)
83. Janmaat KRL, Byrne RW, Zuberbühler K. 2006 89. McNair JN. 1982 Optimal giving-up times and the 96. Barack DL, Ludwig VU, Parodi F, Ahmed N, Brannon
Evidence for a spatial memory of fruiting marginal value theorem. Am. Nat. 119, 511–529. EM, Ramakrishnan A, Platt ML. 2024 Attention
states of rainforest trees in wild mangabeys. Anim. (doi:10.1086/283929) deficits linked with proclivity to explore while
Behav. 72, 797–807. (doi:10.1016/j.anbehav.2005. 90. Van den Driessche C, Chevrier F, Cleeremans A, foraging. Figshare. (doi:10.6084/m9.figshare.c.
12.009) Sackur J. 2019 Lower attentional skills predict 7049958)

You might also like