The Journal of Neuroscience, May 25, 2011 • 31(21):7763–7774 • 7763

Behavioral/Systems/Cognitive

Mechanisms of Rule Acquisition and Rule Following in

Inductive Reasoning
Cristiano Crescentini,1,2 Shima Seyed-Allaei,1 Nicola De Pisapia,3 Jorge Jovicich,3,4 Daniele Amati,1 and Tim Shallice1,4,5
1Cognitive Neuroscience Sector, International School for Advanced Studies, 34136 Trieste, Italy, 2Magnetic Resonance Unit, Azienda Ospedaliero-Universitaria
Santa Maria della Misericordia, 33100 Udine, Italy, 3CIMeC - Center for Mind/Brain Sciences and 4Department of Cognitive Science and Education, University of
Trento, 38068 Rovereto, Italy, and 5Institute of Cognitive Neuroscience, University College London, London WC1N 3AR, United Kingdom

Despite the recent interest in the neuroanatomy of inductive reasoning processes, the regional specificity within prefrontal cortex (PFC)
for the different mechanisms involved in induction tasks remains to be determined. In this study, we used fMRI to investigate the
contribution of PFC regions to rule acquisition (rule search and rule discovery) and rule following. Twenty-six healthy young adult
participants were presented with a series of images of cards, each consisting of a set of circles numbered in sequence with one colored
blue. Participants had to predict the position of the blue circle on the next card. The rules that had to be acquired pertained to the
relationship among succeeding stimuli. Responses given by subjects were categorized in a series of phases either tapping rule acquisition
(responses given up to and including rule discovery) or rule following (correct responses after rule acquisition). Mid-dorsolateral PFC
(mid-DLPFC) was active during rule search and remained active until successful rule acquisition. By contrast, rule following was associ-
ated with activation in temporal, motor, and medial/anterior prefrontal cortex. Moreover, frontopolar cortex (FPC) was active through-
out the rule acquisition and rule following phases before a rule became familiar. We attributed activation in mid-DLPFC to hypothesis
generation and in FPC to integration of multiple separate inferences. The present study provides evidence that brain activation during
inductive reasoning involves a complex network of frontal processes and that different subregions respond during rule acquisition and
rule following phases.

Introduction left DLPFC) in the application and maintenance of rules, rather

Inductive reasoning is central to human learning; it takes specific than in their initial learning. Nonetheless, recent evidence sug-
instances and creates a general rule (Rips, 1999). Such inferences gests that DLPFC plays a role in hypothesis generation during
can be considered to occur in a series of phases. First, the in- rule learning independently of working memory (Boettiger and
stances are collected and held in memory. Second, a regularity is D’Esposito, 2005; Reverberi et al., 2005b; Specht et al., 2009).
identified and a hypothesis generated about relationships be- Frontopolar cortex (FPC) has also been proposed to contrib-
tween the instances. Third, current inferences may be integrated ute to rule learning. However, it is not clear whether it is impor-
with older ones. Finally, generalizations are confirmed through tant in hypothesis generation in general (Strange et al., 2001) or
additional observations. whether it operates in more specific conditions such as when it is
Although it is thought that rule-guided behavior depends on a necessary to integrate the outcomes of more than one inference
network of frontal, parietal, and temporal brain regions (Bunge, (Ramnani and Owen, 2004; Koechlin and Hyafil, 2007). Yet an-
2004; Bunge and Wallis, 2008), the regional specificity for the other possibility is that FPC is involved in maintaining a rule
different mechanisms involved in induction tasks, such as rule active before it becomes familiar; this would fit with its known
learning and rule following, remains underspecified. Thus, some role in intention maintenance (Burgess et al., 2000, 2001) and
studies, including neurophysiological recordings in nonhuman branching (Koechlin et al., 1999).
primates (White and Wise, 1999; Goel and Dolan, 2000; Seger et In this study, we adapted a rule attainment task, the Brixton
al., 2000; Strange et al., 2001; Bunge, 2004) have attributed a role Spatial Anticipation Test (Burgess and Shallice, 1996), to a func-
for the dorsolateral prefrontal cortex (DLPFC) (particularly the tional imaging context with the goal of testing the contribution of
different prefrontal cortex (PFC) and temporo-parietal regions
to rule acquisition and rule following. In this test, participants are
Received Sept. 1, 2010; revised March 22, 2011; accepted March 26, 2011. presented with cards, each consisting of circles numbered se-
Author contributions: C.C., S.S.-A., N.D.P., J.J., D.A., and T.S. designed research; C.C., S.S.-A., and N.D.P. per-
formed research; C.C. and S.S.-A. analyzed data; C.C. wrote the paper.
quentially of which only one is colored blue. The subjects must
This work was partially supported by postdoctoral fellowships awarded to C.C. [International School for Ad- predict the position of the blue circle on the next card. Generally,
vanced Studies (SISSA)] and N.D.P. (Società Scienze Mente Cervello/Fondazione Cassa di Risparmio di Trento e while searching for the correct rule, participants make a series of
Rovereto), and a doctoral fellowship awarded to S.S.-A. (SISSA). incorrect responses before coming up with a correct hypothesis
Correspondence should be addressed to Cristiano Crescentini, International School for Advanced Studies, Cogni-
tive Neuroscience Sector, Via Bonomea, 265, 34136 Trieste, Italy. E-mail: crescent@[Link].
for the current rule. Once correctly acquired, a rule is typically
DOI:10.1523/JNEUROSCI.4579-10.2011 followed perfectly until a new one comes into operation. If
Copyright © 2011 the authors 0270-6474/11/317763-12$15.00/0 DLPFC is involved in hypothesis generation, regardless of the
7764 • J. Neurosci., May 25, 2011 • 31(21):7763–7774 Crescentini et al. • Neural Markers of Rule Acquisition/Rule Following

working memory demands, then this region should be activated justifies separating Acquisition (Search and Discovery phases) from Fol-
consistently until successful rule acquisition. Moreover, we aim lowing parts of the task.
to test the possibility that the FPC should be active during rule Within the Search phase, the response given immediately after a rule
acquisition but that the activation would continue during rule change (SearchPhase1) in which the subject has no useful information
clearly needs to be distinguished from the following responses (Search-
following until any uncertainty concerning the correctness of the
Phase2). Moreover, with respect to the Following phase, the subjects may
rule is resolved. remain uncertain immediately after rule discovery that they have cor-
rectly attained the rule, but this uncertainty will fade over the correct
trials that follow. For this reason, we separated the second correct re-
Materials and Methods sponse in the successful sequence (FollowingPhase1) from the following
Participants. Twenty-six right-handed healthy volunteers (nine females; correct responses (FollowingPhase2). In sum, the key aspect of the study
26.1 ⫾ 4.9 years) participated in the study. All participants had no exist- concerned the categorization of responses in correctly acquired rules into
ing neurological or psychiatric illness. All participants gave written in- five different phases.
formed consent, and the study was approved by the independent Ethics The five identified phases were as follows: Acquisition part: (1) Search-
Committee of the University of Trento. A total of six subjects were ex- Phase1: the first response with a new rule; (2) SearchPhase2: all the re-
cluded from the behavioral and neuroimaging analysis: for one subject, sponses after SearchPhase1 that precede the successful sequence of
technical problems in recording vocal responses precluded the analysis of correct responses; (3) Discovery: the first correct response in the success-
his performance; for the other five excluded participants, the task perfor- ful sequence; Following part: (4) FollowingPhase1: the second correct
mance was too poor (⬍50% of the rules were acquired, with only 23% of response in the successful sequence; (5) FollowingPhase2: the rest of the
the rules being obtained on average). By contrast, all the final sample of correct responses (i.e., from the third) in the successful sequence until the
20 subjects, on which we based all the reported analyses, gave clear vocal rule change.
responses and acquired ⬎50% of the rules, obtaining on average 72.3% This categorization of phases allowed us to contrast the Acquisition
of them. and the Following parts of the task, and subsequently to look more in
Stimuli and design. The Brixton Spatial Rule Attainment Task is a rule detail for activation differences within the Acquisition part (Search-
acquisition and rule following test. In the standard version (Burgess and Phase1 vs SearchPhase2 vs Discovery) and within the Following part
Shallice, 1996), participants are presented one at time with a series of (FollowingPhase1 vs FollowingPhase2).
cards each containing a 2 ⫻ 5 display of circles numbered sequentially. An important issue relates to the type of rules used. As mentioned
One circle only is colored blue, the rest being presented in outline only. above, the total set of 30 rules contained both easy and complex in-
Subjects must predict which circle will be blue on the next card. The rules stances. The difficulty level of rules was evaluated according to minimum
that have to be determined pertain to the relationship between successive description length (MDL) (Solomonoff, 1964; Rissanen, 1978). The idea
stimuli. Typical examples of rules used in the test are the ⫹1 rule (e.g., of MDL is based on Kolmogorov complexity (Li and Vitányi, 1997) that
from 3 to 4, then to 5, and so on) and alternating between circles 5 and 10 measures the amount of information needed to specify an object. For
(for additional examples, see Table 1). In the standard version, each rule example, consider the following two strings of length 14: 1 2 1 2 1 2 1 2 1
is in operation from three to eight trials, and then it changes without 2 1 2 1 2 and 5 8 1 3 2 5 5 6 1 2 2 8 3 6. The first string can simply be
warning. described in English as “1 2 seven times,” whereas the second string is
In the present study, we adapted the standard version of the Brixton generated randomly and the simplest description is the string itself. More
test to a functional magnetic resonance imaging (fMRI) context, the formally, the Kolmogorov complexity of an object is the length of the
main difference being that each card contained a 2 ⫻ 6 display of circles shortest program in any computer programming language that can gen-
(Figs. 1, 2 A). The number of circles in each card was increased from 10 to erate that object.
12 to increase the variability as well as the complexity of the rules. More- The application of Kolmogorov complexity to cognitive science, with
over, in the current version, a card without any filled circles was pre- the aim of measuring difficulty in cognitive tasks, requires that some
sented between each pair of colored cards. This produced a brief rest assumption is made about the types of mental representations involved
period between the responses given by participants. in the task. We assume that there are two kinds of possible representa-
The experimental procedure involved 30 different rules, which ranged tions that can be used to solve the Brixton test. Either or both arithmetic
from very easy ones such as the ⫹1 rule (see above) to more complex and geometric operations can be used. The arithmetic operations are
ones, such as adding 1, then adding 2, then adding 3, and so on. The rules addition and subtraction, whereas in a geometric operation the current
were organized into six fMRI runs of five rules each; in each run, the rules blue circle would move to one of the five neighboring positions. Based on
were presented one after the other. The design was self-paced (for more the kind of operations considered, different MDLs can be assigned to
details, see below, Procedures), so each run had a different duration. On each rule. In general, the difficulty level of each rule was defined as the
average, a run lasted for 6.24 min (⫾0.65; range, 5.26 –9.05 min) and the minimum of the arithmetic and geometric MDLs. As an example, Figure
total time of fMRI scanning was on average 37.4 min (⫾3.1; range, 1 represents two rules that we used in the task together with their MDLs.
33.18 – 44.95 min). The 30 rules we used were divided equally into easy and difficult ones.
In the Brixton test, Search, Discovery, and Following phases can gen- The MDLs varied from 1 to 3 for easy rules and from 3 to 6 for difficult
erally be separated. In fact, the first response to a new rule is typically a ones. In the case of two MDLs of 3, the rule with the shorter mean
guess; this may be followed by a series of incorrect responses, before the arithmetic and geometric MDLs was considered as easier. Table 1 gives all
rule is correctly discovered by subjects; then the rule is followed until the the 30 rules used and, for each, the arithmetic and geometric descrip-
next rule change. In the present study, a rule was judged as correctly tions, the MDL, the difficulty level, the period, and the average discovery
acquired if and only if at least the final two responses were correct (a trial. The period of a given rule denotes the number of cards before the
successful sequence). In other words, a sequence of correct responses rule starts repeating (excluding the first card). The average discovery trial
could not be followed by an incorrect response on the last or next-to-last specifies the trial at which rule acquisition occurs. This corresponds to
cards. the Discovery phase, namely when the first correct response occurs in a
To assess whether a separation between Search, Discovery, and Fol- successful sequence of correct responses. It naturally follows that, on
lowing phases also applied to the rule set used in this study, we considered average, easy rules have a shorter period than difficult rules (1.66 vs 3
all rules on which the participants were correct on the last two responses. trials; t(27) ⫽ ⫺5.53; p ⬍ 0.01) and are acquired more quickly (after 3.59
In a high proportion of them, namely 74%, the subjects were consistently vs 5.10 trials; t(28) ⫽ ⫺4.21; p ⬍ 0.01). As shown in the table, both types
wrong and then consistently right. In the remaining 26% of the rules, the of rules are typically acquired after the completion of the corresponding
subjects produced a correct response in the Search phase (i.e., a correct period.
response followed by an incorrect one before the successful sequence). Distinguishing between easy and difficult rules was important since, in
Most of these correct responses were guesses after a rule change. This a subsidiary investigation, we examined whether there were brain regions
Crescentini et al. • Neural Markers of Rule Acquisition/Rule Following J. Neurosci., May 25, 2011 • 31(21):7763–7774 • 7765

Table 1. The 30 rules used during fMRI

Rule ID Rule Example Arithmetic description Geometric description ADL GML Per MDL ADT DL

1 Same 4-4-4-4-4-4 Same Same 1 1 1 1 2.72 E

2 Add 2 8-10-12-2-46 ⫹2 33 3 2 1 2 2.89 E
3 Subtract 2 8-6-4-2-12-10 ⫺2 44 3 2 1 2 2.72 E
4 ⫹1, 0 6-7-7-8-8 ⫹1 3 3 2 2 2 3.82 E
0 .
5 ⫺1, 0 6-5-5-4-4-3-3 ⫺1 4 3 2 2 2 4.21 E
0 .
6 Top down 1-8-2-9-3-10 ⫹7 n 8 2 2 2 2.71 E
⫺6 1
7 Bottom up 11-4-10-3-9-2 ⫺7 b 8 2 2 2 3.33 E
⫹6 2
8 Diagonal 2-9-4-11-6-7 ⫹7 n 8 2 2 2 3.57 E
⫺5 m
9 Add 3 3-6-9-12-3 ⫹3 333 3 3 1 3 2.94 E
10 Subtract 3 2-11-8-5-2 ⫺3 444 3 3 1 3 3 E
11 ⫹2, 0 6-8-8-10-10-12-12 ⫹2 33 4 3 2 3 3.89 E
0 .
12 ⫺2, 0 8-6-6-4-4-2-2 ⫺2 44 4 3 2 3 4.76 E
0 .
13 1; same 1-6-2-6-3-6-4-6 x ⫽ current pos, pos x ⫽ current pos, pos 3 3 2 3 5.37 E
x⫹1 x3
14 ⫹2, ⫹1 9-11-12-2-3-5-6 ⫹2 33 5 3 2 3 3.82 E
⫹1 3
15 ⫺2, ⫺1 9-7-6-4-3-2-1 ⫺2 44 5 3 2 3 4.14 E
⫺1 4
16 ⫹2, ⫺1 5-7-6-8-7-9-8 ⫹2 33 5 3 2 3 4.66 D
⫺1 4
17 ⫺2, ⫹1 4-2-3-1-2-12 ⫺2 44 5 3 2 3 3.91 D
⫹1 3
18 Vertical 12-6-10-4-8-2-12 ⫺6 1 8 3 2 3 4.91 D
⫹4 ,4
19 Parallel diagonal 5-12-4-11-3-10 ⫹7 n 9 3 2 3 3.94 D
⫺8 4b
20 Left loop diagonal 6-5-12-5-4-11-4 ⫺1 4 10 3 3 3 6.5 D
⫹7 n
⫺7 b
21 Right loop diagonal 1-2-9-2-3-10-3-4 ⫹1 3 10 3 3 3 5.90 D
⫹7 n
⫺7 b
22 1 (2), 2 6-7-8-10-11-12-2 ⫹1 (2) 3 6 4 3 4 6.16 D
⫹2 3
33
23 Greek fret 11-5-4-10-9-3 ⫹6 1 6 4 4 4 3.93 D
⫹1 3
2
3
24 Extremes 12-6-1-7-12-6-1-7 ⫺6 1 8 4 4 4 3.89 D
⫺5 3
2
4
25 Central square 9-3-4-10-9-3-4-10 ⫺6 1 12 4 4 4 4.37 D
⫹1 3
⫺6 2
⫺1 4
26 Series on two rows 1-6-2-7-3-8-4-9-5-10-6 ⫹5 4 (only once) 8 4 3 4 4.78 D
⫺4 33 (only once)
2
m
27 1; diagonal 8-9-4-5-12-7 ⫹1 3 12 4 4 4 6.69 D
⫺5 m
⫹1 3
⫹7 n
28 ⫹1, ⫹2 (2) 3-4-6-8-9-11-13 ⫹1 3 6 4 3 4 5.92 D
⫹2 repeat 33 repeat
29 3, 1 (2) 7-10-11-12-3-4-5-8 ⫹3 333 6 5 3 5 6.91 D
⫹1 repeat 3
3
30 Progression 7-8-10-1-5-10-4 Load x, ⫹ x — 6 45 — 6 4.00 D
x⫽x⫹1
The numbers in the example column refer to the position of the blue circle in the 2 row ⫻ 6 column array. ADL, Arithmetic description length; GML, geometric description length; Per, period (number of cards, excluded the first, before rule
starts repeating); MDL, minimum description length; ADT, average discovery trial (trial at which the rule is acquired); DL, difficulty level (easy/difficult). Despite equal ADL, GML, Per, and MDL, rules 16 and 17 were considered to be more
difficult than rules 14 and 15 because they involve two different arithmetic operations.
7766 • J. Neurosci., May 25, 2011 • 31(21):7763–7774 Crescentini et al. • Neural Markers of Rule Acquisition/Rule Following

modulated by rule difficulty in the Acquisition

versus the Following part of the task. With re-
spect to this, we first tested for the main effect
of difficult minus easy rules, considering all
rule phases together. We then applied an exclu-
sive masking procedure, using the main effect
as the masking contrast, to identify voxels in
which any potential rule difficulty effect is not
shared between the two contrasts of interest:
(1) difficult minus easy rules computed for the
rule acquisition part and (2) the corresponding
contrast performed for the rule following part.
The categorization of responses in different
phases, as explained above for the overall set of
rules, remained the same for easy and difficult
rules. However, there was an exception: re-
sponses given after a rule change (i.e., Search-
Phase1) were not analyzed separately for the
two types of rules since all rules can be consid-
ered equivalent at this stage.
Procedures. Figure 2 A shows a schematic
representation of the experimental paradigm
during fMRI. This comprised 30 rules orga- Figure 1. Examples of two rules in the Brixton test. Five cards are presented for each rule. Each rule can be described either in
nized in six runs. Each run consisted of five geometric or arithmetic terms. Geometrically, the first rule (rule 1: top down) (see also rule ID 6 in Table 1) can be described as “right
random rules balanced for difficulty and num- down diagonal,” then “up.” If we consider moving to each neighbor of a position to be a geometric operation of length 1, then the
ber of cards within runs. Each rule occurred geometric MDL is 2 for this rule. Arithmetically, this rule can be described as “⫹7,” then “⫺6.” To allow the arithmetic description
only once in the experiment and consisted of a length to consider smaller operands like ⫹1 easier than larger operands like ⫹7, the description length of operands is given by
variable number of colored cards. A rule was in their length in the binary system. The description length of each arithmetic operator is also assumed to be 1. In the example, both
operation for a minimum of 6 colored cards “⫹7” and “⫺6” have an arithmetic length of 3 ⫹ 1 ⫽ 4, and the total arithmetic MDL for the rule is 4 ⫹ 4 ⫽ 8. Thus, the
and a maximum of 10 colored cards; easier minimum of arithmetic and geometric MDLs is 2 for this rule. The second rule (rule 2: add 3) (see also rule ID 9 in Table 1) can be
rules consisted of fewer cards than more diffi- described geometrically as “right right right” and arithmetically as “⫹3,” which has MDL of 3 in both cases.
cult rules (on average 6.8 cards vs 9.2 cards,
respectively). the scanning session, anatomical scans were also acquired for each par-
Each colored card was presented for a maximum of 7 s. Participants ticipant, using a T1-weighted MP-RAGE.
were instructed to use this time to think about the circle that will appear The experimental task was presented using ASF (“A Simple Frame-
colored blue on the next card. Subjects were asked to press a button when work”) (available at [Link] based on Psychtool-
ready to give a response. At the button press, the colored card was re- box for MATLAB (Brainard, 1997). SPM8 (Wellcome Department of
placed by an empty card and subjects were required to say aloud the Cognitive Neurology; see also [Link]
number where the next blue circle would appear as defined by the current ware/spm8/) was used for data preprocessing and statistical analyses. For
rule. Subjects were told to consider the button press as a way to switch on all participants, we acquired on average 1020 volumes (170 volumes on
a microphone. Vocal responses were collected for a maximum of 4 s after average for each fMRI run); the first 4 volumes were discarded for each
a button press. After this time and a jitter of 1–3 s, the next colored card run to allow the magnetization to reach steady state. Slice acquisition
was presented. Subjects were told that a rule could change after a certain delays were corrected using the middle slice as reference. All images were
number of colored cards and that they would then need to detect the then corrected for head movement. All images were normalized to the
change and infer the new rule as rapidly as they could. They were dis- standard SPM EPI template and spatially smoothed using an 8 mm
couraged from trying to predict when the rule would change but to FWHM Gaussian filter. The high-pass filter was set to the cutoff value of
follow the learned rule as long as it was in operation. Fixation blocks of 128 s.
15 s each were added both at the beginning and at the end of each run. All subsequent analyses of the functional images were performed using
During fMRI, overt verbal responses were acquired using an MR- the general linear model implemented in SPM8. First, for each subject,
compatible fiber optic microphone that was attached to the head coil the data were fitted at every voxel using a combination of effects of
(Fibersound; model FOM1-MR); each participant’s response was re- interest. For correctly acquired rules, we modeled the onset of each col-
corded as a .wav file. The responses were transcribed and checked for ored card separately for all the five phases (i.e., SearchPhases1&2, Dis-
accuracy. For all subjects included in the analysis, all responses were covery, FollowingPhases1&2) and convolved with the hemodynamic
recorded clearly. response function (HRF). Each phase was modeled using a boxcar with a
Before fMRI scanning, participants practiced the tasks for ⬃7 min. variable duration, so that the resultant ␤ values represented an estimate
Subjects were first explained the test and then presented with five rules of the neural response per unit time spent thinking about the future
using the same procedure to the one used in the scanner. The rules used position of the blue circle. The duration of each boxcar function corre-
during practice were different from those used during fMRI. Moreover, sponded to the latency of the response (button press) generated by the
during practice, subjects were instructed on how to produce spoken subject in the corresponding phase.
responses while moving the face muscles as little as possible. On average, the block duration of the different phases was as follows:
fMRI methods: acquisition and analyses. Images were acquired using a 4 2.98 s (SearchPhase1), 2.76 s (SearchPhase2), 2.60 s (Discovery3), 1.49 s
T MRI scanner (Bruker Medical) using a birdcage transmit, eight- (FollowingPhase1), and 1.26 s (FollowingPhase2) (see also Fig. 2 B). Al-
channel receive head coil (USA Instruments) and the point spread func- though there was no difference between the block duration for the first
tion distortion-corrected echoplanar imaging (Zaitsev et al., 2004). Head three phases (all p ⬎ 0.09) (see Results, Behavioral data), these blocks
movement was minimized by mild restraint and cushioning. Thirty- lasted significantly longer than the blocks in FollowingPhases1&2. More-
seven slices of functional MR images were acquired using blood oxygen- over, the block duration of FollowingPhase1 was longer than that of
ation level-dependent imaging (3 ⫻ 3 mm; 3 mm thick; repetition time, FollowingPhase2 ( p ⬍ 0.001). Considering these block durations, for
2.2 s; time echo, 21 ms), covering the entire cortex. At the beginning of each subject and for each of the five phases we acquired on average 17.71
Crescentini et al. • Neural Markers of Rule Acquisition/Rule Following J. Neurosci., May 25, 2011 • 31(21):7763–7774 • 7767

and rule use (Donohue et al., 2005; Badre et al.,

2010). Given our hypotheses and the findings
of whole-brain analysis, the first two ROIs were
defined in left mid-DLPFC (⫺50 26 24) and
left FPC (⫺36 50 6). These regions were de-
rived from the study by Badre et al. (2010).
These regions were also anatomically close to
those reported in other functional imaging
studies of reasoning [Hampshire et al. (2011),
their Table 4]. A third ROI was centered in the
superior parietal lobule (SPL) (⫺27 ⫺52 57)
and was again based on the study by Badre et al.
(2010). A fourth ROI was defined in the poste-
Figure 2. A, Experimental paradigm. The timing of the succession of the events is reported. Each given rule consists of a variable rior middle temporal gyrus (postMTG) (⫺56
number of colored cards (in which a circle is colored blue) and empty cards in which all circles are presented white. When presented ⫺40 2); as no activation in this region was re-
with colored cards, subjects are required to press a button when ready to predict which circle will be colored blue in the next colored ported in the study by Badre et al. (2010), al-
card. Immediately after button press, colored cards are replaced with empty cards. During the presentation of empty cards, though this region has been associated with
participants are asked to say aloud the number of the corresponding circle. The next colored card follows an empty card after a rule storage and rule use in past research
variable time (1–3 s). B, Behavioral results during fMRI. The graph reports mean RT of responses (button press) given in the five (Bunge, 2004), the coordinates were derived
phases identified in correctly acquired rules. Phase connotation is as follows: (1) SearchPhase1 (SrP1) denotes the first response from the study by Donohue et al. (2005), a key
with a new rule, (2) SearchPhase2 (SrP2) identifies all the responses after SearchPhase1 that precede a successful sequence of paper with respect to the role of the posterior
correct responses, (3) Discovery (Dsv) corresponds to the first correct response in the successful sequence, (4) FollowingPhase1 temporal cortex in rule use. Since the whole-
(FlP1) denotes the second correct response in the successful sequence, and (5) FollowingPhase2 (FlP2) corresponds to the rest of brain analysis did not produce lateralization ef-
correct responses (from the third) in the successful sequence until the rule change. Error bars indicate SD. fects, we also used the homologues regions in
the right hemisphere [for the right SPL, the
coordinates were 24 ⫺56 55, and were derived
volumes for SearchPhase1, 62.75 for SearchPhase2, 24.98 for Discovery, from the study by Badre et al., (2010)].
13.74 for FollowingPhase1, and 29.28 for FollowingPhase2. The ROIs were constructed by creating a sphere with a radius of 6 mm
In addition, the model comprised the onset of each fixation period around the centers defined by the sets of coordinates reported above. The
(block duration, 15 s) and also, as two separated regressors, the onsets of mean ROI data for each event type (five phases plus fixation baseline
each colored card presented for the rules that were not acquired [block blocks) were obtained from the design matrix of each subject. Using the
duration dependent on mean reaction time (RT); mean duration equal to Marsbar toolbox (Brett et al., 2002), the percentage signal change of all
2.53 s], plus the onset of the vocal responses (block duration, 3 s), con- phases, including fixation baseline blocks, for each subject’s ROI were
volved with the HRF. The latter two regressors were excluded from computed for an event duration of 0 s and the absolute maximum func-
subsequent group-level analyses. Finally, the first-level analyses also tion was used to calculate the signal change from each canonical event.
included the parameters of the realignment (motion correction) as The resulting data for all six runs were averaged over runs and subjects.
covariates of no interest. All ROI data were used in repeated-measures ANOVAs.
Next, we obtained five contrast images per participant, corresponding
to the five conditions of interest (the five phases) versus the baseline Results
period (i.e., fixation) and pooling across the six runs. These five contrasts Behavioral data
were then submitted to a 1 ⫻ 5 full factorial ANOVA, for group-level
Participants acquired on average 21.7 rules (⫾4.19; range, 16 –
random effects statistical inference. Of importance, to test for the average
effects for contrasts between rule acquisition and rule following that are 29). Subjects also acquired more easy than difficult rules (11.9 ⫾
unbalanced (i.e., SearchPhases1&2 and Discovery minus Following- 2.17 vs 9.5 ⫾ 2.56; t(19) ⫽ 4.90; p ⬍ 0.001).
Phases1&2 and reversed), the contrast vectors were balanced by being The RT data for each of the five phases identified in correctly
defined as [1/3 1/3 1/3 ⫺1/2 ⫺1/2] and [⫺1/3 ⫺1/3 ⫺1/3 1/2 1/2], acquired rules are shown in Figure 2 B. A 1 ⫻ 5 repeated-measure
respectively. ANOVA with the factor of rule phase (from SearchPhase1 to
Moreover, in a separate ANOVA, we tested for effects of rule difficulty. FollowingPhase2) revealed differences in RT between the differ-
This was done by specifying a model similar to that described above with ent phases (F(1,19) ⫽ 40.53; p ⬍ 0.001). Pairwise comparisons
the only exception being that this additional model included easy and assessed the significance of the magnitude of the differences.
difficult correctly acquired rules, separately. Correction for nonsphericity These showed that rule acquisition, namely SearchPhases1&2
(Friston et al., 2002) was used in both ANOVAs to account for possible
and Discovery, was associated with slower RT than rule following
differences in error variance across conditions and any nonindependent
error terms for the repeated measures. Statistical threshold was set to (FollowingPhases1&2). In fact, the responses in all three of the
pcorr ⫽ 0.05 corrected for multiple comparisons at the cluster level (clus- acquisition phases were slower than those produced during the
ter size estimated at punc ⫽ 0.001), considering the whole brain as the two rule following phases (all F ⬎ 32.91; all p ⬍ 0.001). Of inter-
volume of interest. Furthermore, with respect to the ANOVA performed est, although there were no differences between the three rule
to assess rule difficulty effects, the SPM constituting the exclusive mask acquisition phases (all F ⬍ 3.08; all p ⬎ 0.09), the second given
was thresholded at the default p ⬍ 0.05, whereas the contrasts to be responses in a successful sequence of correct responses (i.e., Fol-
masked were thresholded at p ⬍ 0.001 and then corrected for multiple lowingPhase1) were slower than those produced subsequently
comparisons at the cluster level ( pcorr ⫽ 0.05). (i.e., FollowingPhase2; p ⬍ 0.001).
Finally, the whole-brain analyses aimed to assess the effects of rule
acquisition versus rule following were supplemented with region of in-
Neuroimaging data
terest (ROI) analyses. An ROI approach was used to confirm effects
found in the whole-brain corrected data and, more critically, to test for The main aim of the fMRI analyses was to investigate whether
theoretically relevant dissociations in activation between regions and there exist regions preferentially active during rule acquisition
conditions (i.e., the five rule phases) (Henson, 2005). Location of ROIs and also others contributing more to rule following. Accordingly,
was unbiased with respect to the tests of interest since the coordinates we first report the results relative to the contrasts aimed to high-
were taken from two previous imaging studies of abstract rule learning light brain regions involved in acquiring rules, and next we will
7768 • J. Neurosci., May 25, 2011 • 31(21):7763–7774 Crescentini et al. • Neural Markers of Rule Acquisition/Rule Following

Table 2. Rule acquisition minus rule following

MNI coordinates
Anatomical localization ⬃BA x y z pcorr Z value Voxels per cluster
Main effect: regions more active for SearchPhases1&2 and Discovery than FollowingPhases1&2
L middle frontal gyrus 6 ⫺40 4 40 ⬍0.001 6.20 2795
L middle frontal gyrus 9 ⫺46 28 36 4.72
L middle frontal gyrus 46 ⫺44 22 26 4.55
L superior frontal gyrus 6/8 ⫺28 0 68 3.60
L inferior frontal gyrus 45/46 ⫺54 30 18 3.37
R middle frontal gyrus 6 30 ⫺2 54 ⬍0.001 5.15 1923
R middle frontal gyrus 9 50 24 30 4.56
R superior parietal lobe 7 34 ⫺46 48 ⬍0.001 4.21 789
R inferior parietal lobe 40 42 ⫺36 48 3.90
R precuneus 19 30 ⫺66 42 3.74
L superior parietal lobe 7 ⫺30 ⫺48 48 ⬍0.02 3.77 405
L precuneus 7 ⫺26 ⫺68 36 3.33
Conjunction: regions more active for SearchPhase1 minus FollowingPhases1&2 艚
SearchPhase2 minus FollowingPhases1&2 艚 Discovery minus FollowingPhases1&2
L middle frontal gyrus 6 ⫺40 2 42 ⬍0.001 5.02 1267
L middle frontal gyrus 9 ⫺48 26 30 3.80
L middle frontal gyrus 8 ⫺50 20 38 3.60
R middle frontal gyrus 6 28 ⫺4 56 ⬍0.002 4.49 655
R middle frontal gyrus 9 54 22 30 3.75
Stereotactic MNI coordinates for significant clusters (random effects, cluster-level p ⬍ 0.05, corrected, estimated at p ⬍ 0.001, uncorrected) given in millimeters with effect sizes (z scores) and cluster extent. In the voxels per cluster column,
cluster extent is reported in correspondence of the main peak. Subpeaks were selected dividing each cluster into Brodmann areas (BA) and then selecting peaks within each area. L, Left; R, right.

turn to patterns of activation related to

rule following. For rule acquisition, we
first focus on the main effect of Search-
Phases1&2 and Discovery minus Follow-
ingPhases1&2 (Table 2, first contrast; Fig.
3). For rule following, we focus on the
main effect of FollowingPhases1&2 minus
SearchPhases1&2 and Discovery (Table 3, Fig.
4). Moreover, repeated-measures ANOVAs
were performed on ROI data to test for
significant interactions between region
and condition (i.e., the five phases identi-
fied in correctly acquired rules).
As already mentioned, we also investi-
gated whether rule difficulty modulated
activity in any brain region, either during
rule acquisition or rule following. Accord-
ingly, we first report the main effect of dif-
ficult minus easy rules (Table 4, first
contrast), and then we focus on the two
corresponding simple main effects, which
were computed separately for the acquisi-
tion phases and the following phases using
the main effect as an exclusive mask (Ta- Figure 3. Effect of rule acquisition minus rule following (i.e., SearchPhases1&2 and Discovery minus FollowingPhases1&2). Brain
ble 4, second contrast). [Link]
[Link][inarbitraryunits(a.u.);⫾90%
Rule acquisition versus rule following confidenceinterval];plotsreportthepatternofactivityatthepeaksofactivation(i.e.,singlevoxels)selectedfromthewhole-braincontrast
For rule acquisition minus rule following, SPM maps. The bottom part of the figure reports brain activation in both hemispheres for the conjunction of rule acquisition versus
we compared activation during Search- combined rule following simple effects. The analysis was performed to test for any common effect of rule acquisition ⬎ rule following,
Phases1&2 and Discovery versus Follow- regardless of the individual acquisition phases. The conjunction is given by the following: SearchPhase1 minus FollowingPhases1&2 艚
ingPhases1&2. This contrast revealed SearchPhase2minusFollowingPhases1&2艚DiscoveryminusFollowingPhases1&2.SrP1,SrP2,Dsv,FlP1,andFlP2refertoSearchPhase1,
activation for rule acquisition ⬎ rule fol- SearchPhase2, Discovery, FollowingPhase1, and FollowingPhase2, respectively.
lowing in four clusters (Table 2, first con-
trast). The largest cluster was a swathe of also activated in this cluster. A similar pattern of activation in
cortex in the left hemisphere from the mid-DLPFC through the the corresponding regions of the right hemisphere was found
inferior frontal junction area (IFJ) (Brass et al., 2005). This latter in the second largest cluster; activation in this cluster extended
region was very close to what is termed the dorsal anterior pre- slightly more posteriorly in the dorsal premotor cortex (PMd)
motor cortex (pre-PMd) in the studies by Badre and D’Esposito (Badre et al., 2010). Moreover, the superior parietal lobe and
(2007) and Badre et al. (2010). The left superior frontal gyrus was the precuneus were activated in both hemispheres as part of
Crescentini et al. • Neural Markers of Rule Acquisition/Rule Following J. Neurosci., May 25, 2011 • 31(21):7763–7774 • 7769

Table 3. Rule following minus rule acquisition

MNI coordinates
Anatomical localization ⬃BA x y z pcorr Z value Voxels per cluster
Main effect: regions more active for FollowingPhases1&2 than SearchPhases1&2 and Discovery
L superior temporal gyrus 41 ⫺56 ⫺26 10 ⬍0.001 6.54 11,770
L superior temporal gyrus 22 ⫺56 ⫺12 6 6.27
L precentral gyrus 4 ⫺58 ⫺4 18 6.20
L precentral gyrus 6 ⫺50 ⫺14 32 5.66
L postcentral gyrus 3 ⫺24 ⫺28 62 5.59
R precentral gyrus 4 18 ⫺24 66 5.42
L parahippocampal gyrus // ⫺26 ⫺2 ⫺14 5.39
L cingulate gyrus 24 ⫺4 0 44 5.34
L insula 13 ⫺34 0 12 5.17
L putamen // ⫺28 ⫺14 2 4.86
L inferior parietal lobe 40 ⫺64 ⫺26 26 4.37
L postcentral gyrus 5 ⫺20 ⫺40 68 4.23
R superior temporal gyrus 22 52 ⫺2 0 ⬍0.001 6.51 6187
R superior temporal gyrus 21 64 ⫺14 ⫺2 6.36
R superior temporal gyrus 41 50 ⫺28 10 6.12
R precentral gyrus 6 64 2 10 6.09
R parahippocampal gyrus // 26 ⫺2 ⫺14 4.70
R putamen // 28 ⫺10 ⫺2 4.23
R inferior parietal lobe 40 66 ⫺28 28 3.28
R inferior frontal gyrus 47 58 22 0 3.16
R postcentral gyrus 2 68 ⫺22 32 3.14
R cerebellum // 18 ⫺58 ⫺22 ⬍0.001 6.50 3637
L cerebellum // ⫺18 ⫺62 ⫺22 4.80
L cuneus 19 ⫺10 ⫺84 30 4.63
R cuneus 19 8 ⫺80 28 4.57
R lingual gyrus 18 16 ⫺62 ⫺2 4.30
L posterior cingulate 31 ⫺2 ⫺68 16 3.35
R posterior cingulate 31 2 ⫺70 14 3.28
R medial frontal gyrus 10 10 56 8 ⬍0.001 5.21 1130
L medial frontal gyrus 10 ⫺10 52 10 4.66
R anterior cingulate 32 6 38 ⫺4 3.89
L anterior cingulate 32 ⫺6 38 0 3.64
Stereotactic MNI coordinates for significant clusters (random effects, cluster-level p ⬍ 0.05, corrected, estimated at p ⬍ 0.001, uncorrected) given in millimeters with effect sizes (z scores) and cluster extent. In the voxels per cluster column,
cluster extent is reported in correspondence of the main peak. Subpeaks were selected dividing each cluster into Brodmann areas (BA) and then selecting peaks within each area.

two separate clusters. The activation also extended to the in- Moreover, because we expected a similar effect of rule acqui-
ferior parietal lobe in the right hemisphere. Of interest, the sition ⬎ rule following in the DLPFC regardless of the different
FPC did not show an effect of rule acquisition minus rule phases, we formally assessed this prediction by performing a con-
following. Figure 3 shows the areas that were activated for the junction analysis (Nichols et al., 2005) of SearchPhase1 minus
main effect, plus the signal plots for the left and right mid- FollowingPhases1&2 艚 SearchPhase2 minus FollowingPhases1&2
DLPFC and parietal cortex, including all five phases. The plots 艚 Discovery minus FollowingPhases1&2. The bottom part of
demonstrate an effect of rule acquisition minus rule following Figure 3 shows that the only regions that were significantly and
(SrP1, SrP2, and Dsv minus FlP1 and FlP2 in the plots) in all consistently more active during each phase of rule acquisition
four regions. Nevertheless, whereas the mid-DLPFC appeared than during rule following were the mid-DLPFC and a more
to be consistently more activated in both hemispheres during posterior region that includes the IFJ/pre-PMd area. Both of
SearchPhases1&2 and Discovery than FollowingPhases1&2, these regions were activated bilaterally. The conjunction analysis
the decrease in activation observed in FollowingPhase1&2 was
confirmed the pattern of activation observed in the signal plots
more gradual for the parietal lobe. Consistent with this, Dis-
relative to the right and left mid-DLPFC. Consistent with this, no
covery and FollowingPhase1 trials differed in terms of mid-
activation was found in the mid-DLPFC (or in the parietal lobe,
DLPFC activation (xyz, ⫺42 22 24, z ⫽ 4.43, p ⬍ 0.001; xyz, 46
the IFJ/pre-PMd area, and the FPC) when the three rule acquisi-
24 26, z ⫽ 4.53, p ⬍ 0.001) but not in terms of parietal
activation. tion phases were directly contrasted against each other.
Interestingly, the left mid-DLPFC (xyz, ⫺52 20 28; z ⫽ 4.00; Summarizing, the imaging results for the contrasts involving
p ⬍ 0.001) and the left IFJ (xyz, ⫺40 2 42; z ⫽ 5.77; p ⬍ 0.001) rule acquisition minus rule following phases highlighted the role
were the only active regions, in a single cluster of 1197 voxels of a network including frontal and parietal brain regions in rule
extent, when we tested for the same effect of rule acquisition ⬎ acquisition. More specifically, the mid-DLPFC and IFJ/pre-PMd
rule following, in an ANOVA containing RT as a covariate for regions were consistently activated in both hemispheres in all
each trial. This indicates a modulation of these regions over- phases of rule acquisition and rapidly declined during rule fol-
and-above the effect of RT in our study. This cluster even lowing phases. Unlike the mid-DLPFC, the parietal lobes ap-
includes (xyz, ⫺52 12 32) the peak of the region found by Goel peared to remain active during FollowingPhase1 trials. Finally,
and Dolan (2000) showing an effect of rule application ⬎ the FPC was not part of the regions showing a main effect of rule
perceptual baseline. acquisition minus rule following.
7770 • J. Neurosci., May 25, 2011 • 31(21):7763–7774 Crescentini et al. • Neural Markers of Rule Acquisition/Rule Following

Rule following versus rule acquisition

In our analyses, we also investigated
whether there were regions contributing
more to rule following than rule acquisi-
tion. With respect to this, we tested for the
main effect of rule following minus rule
acquisition, namely FollowingPhases1&2
minus SearchPhases1&2 and Discovery.
This contrast revealed activation in four
large clusters (Table 3). The largest had its
peak of activation in the left superior tem-
poral gyrus. Activation in this cluster also
extended to the left precentral and post-
central gyri as well as to the cingulate
gyrus and to the most lateral part of the
Figure 4. Effect of rule following minus rule acquisition (i.e., FollowingPhases1&2 minus SearchPhases1&2 and Discovery).
inferior parietal lobe. Activation in sub- Brain activation in both hemispheres is reported in the top part of the figure. Signal plots for each rule phase are reported for both
cortical regions such as the left putamen the right and left superior temporal gyrus and the medial prefrontal cortex. Plots depict activity in experimental conditions relative
was also found in this cluster. Moreover, a to fixation [in arbitrary units (a.u.); ⫾90% confidence interval]; plots report the pattern of activity at the peaks of activation (i.e.,
second large cluster involved very similar single voxels) selected from the whole-brain contrast SPM maps. SrP1, SrP2, Dsv, FlP1, and FlP2 refer to SearchPhase1, Search-
regions in the right hemisphere. A third Phase2, Discovery, FollowingPhase1, and FollowingPhase2, respectively.
cluster was centered in the cerebellum and
extended to other posterior brain regions noted that all ROIs, except the right mid-DLPFC one, showed a
such as the cuneus and the lingual gyrus. A fourth cluster in- significant effect of all phases (i.e., from SearchPhase1 to Follow-
volved the anterior, ventral part of the medial frontal gyrus and ingPhase2) minus fixation (all p ⬍ 0.05; p ⫽ 0.18 for the right
extended to the anterior cingulate cortex. Figure 4 shows the mid-DLPFC ROI). This was consistent with the whole-brain
areas that activated for the main effect of rule following minus data, which showed the same effect in voxels within 6 mm of the
rule acquisition, plus the signal plots for the right and left supe- center of each ROI, this time also in the right mid-DLPFC ROI.
rior temporal gyri and for the medial frontal gyrus including all The ROI data were first submitted to a 2 (hemisphere: left and
five phases. The plots demonstrate that there are consistently right) ⫻ 4 (region: mid-DLPFC, FPC, SPL, and postMTG) ⫻ 5
greater activations in these regions during FollowingPhases1&2 (condition: from SearchPhase1 to FollowingPhase2) repeated-
than the rule acquisition phases. Of interest, the plot concerning measures ANOVA. The analysis showed the significant main ef-
the medial frontal gyrus shows that SearchPhases1&2 and Dis- fect of region (F(3,57) ⫽ 10.20; p ⬍ 0.001) but not the effects of
covery activated this region less than fixation. Remarkably, we hemisphere (F(1,19) ⫽ 2.96; p ⫽ 0.101) or condition (F(4,76) ⫽
also found that, during Discovery trials, there was a positive cor- 2.15; p ⫽ 0.082). Post hoc tests (Bonferroni correction applied)
relation between the mean activation in the cluster involving the performed for the region factor showed that the postMTG was
medial PFC (computed using the Marsbar toolbox) and RT in more active than the other three regions (all p ⬍ 0.03) with no
those trials (r ⫽ ⫹0.49; p ⬍ 0.03). This correlation indicated that difference among the three (all p ⬎ 0.11). The analysis also
the less active the areas involved in this cluster were during the showed the two-factor interactions between hemisphere and re-
Discovery phase, the faster participants responded in this phase. gion (F(3,57) ⫽ 7.46; p ⬍ 0.001) and hemisphere and condition
The main effect of rule following minus rule acquisition was (F(4,76) ⫽ 4.50; p ⬍ 0.004). Post hoc tests (Bonferroni correction
also tested in an ANOVA involving RT as a covariate for each applied) executed for the first of the two interactions showed that
trial. In a similar fashion to the basic ANOVA results just de- the mid-DLPFC was more active in the left than the right hemi-
scribed, this main effect revealed bilateral activation in both the sphere ( p ⬍ 0.018) with no difference between the two hemi-
precentral gyrus and the superior temporal gyrus (peak activa- spheres being present for the other three regions (all p ⬎ 0.28).
tion in the right hemisphere cluster: xyz, 58 ⫺8 2; z ⫽ 6.80; p ⬍ Post hoc tests (Bonferroni correction applied) performed for the
0.001; cluster extent, 2726 voxels; peak activation in the left hemi- hemisphere by condition interaction revealed a difference be-
sphere cluster: xyz ⫺56 ⫺28 10; z ⫽ 6.77; p ⬍ 0.001; cluster tween the two hemispheres (left ⬎ right) for the discovery phase
extent, 3728 voxels). Finally, we also assessed whether there were ( p ⬍ 0.015) but not for all other phases (all p ⬎ 0.17). Critically,
differences between the two rule following phases. The only region the ANOVA also showed a significant interaction between region
reporting a significant effect was the left FPC (⫺46 48 0, Z ⫽ 4.80, and condition (F(12,228) ⫽ 11.04; p ⬍ 0.001). This interaction was
pFWE-corr ⫽ 0.021), which was more active for FollowingPhase1 than not modulated by the hemisphere factor (i.e., hemisphere by re-
FollowingPhase2 trials. gion by condition interaction: F(12,228) ⫽ 1.02, p ⫽ 0.43). In view
of this latter result, we averaged the signal across the right and left
Analysis of ROI data hemispheres in each ROI and used this averaged signal in all
An ROI approach was adopted to further characterize the activa- subsequent analyses concerning the mid-DLPFC, FPC, SPL, and
tion profiles observed in prefrontal, temporal, and parietal re- postMTG ROIs. The averaged signal change for the four resulting
gions during rule acquisition and rule following phases. As ROIs is shown in Figure 5 for each phase identified in our test and
already mentioned, we defined ROIs in mid-DLPFC (⫺50 26 24; for the fixation baseline blocks.
50 26 24), FPC (⫺36 50 6; 36 50 6), SPL (⫺27 ⫺52 57; 24 ⫺56 55), To examine where the original region by condition interac-
and postMTG (⫺56 ⫺40 2; 56 ⫺40 2). Critically, this ROI ap- tion arose, we performed six 2 (region) ⫻ 5 (condition) ANOVAs
proach allowed one to test for dissociations between these regions using a Bonferroni correction criterion of p ⬍ 0.0083. In four of
(i.e., region by condition interaction; the factor condition being the ANOVAs, the region by condition interaction was significant,
given by the five phases identified in our test). First, it should be including all the interactions involving the postMTG (postMTG
Crescentini et al. • Neural Markers of Rule Acquisition/Rule Following J. Neurosci., May 25, 2011 • 31(21):7763–7774 • 7771

Table 4. Effect of rule difficulty

MNI coordinates
Anatomical localization ⬃BA x y z pcorr Z value Voxels per cluster
Main effect: regions more active for difficult rules than easy rules
R superior temporal gyrus 22 68 ⫺22 2 ⬍0.001 6.99 7624
R precentral gyrus 6 64 6 10 6.57
R superior temporal gyrus 21 62 ⫺16 ⫺2 6.48
R precentral gyrus 4 54 ⫺4 46 5.83
R precentral gyrus 44 60 14 4 5.60
R inferior frontal gyrus 45 48 28 4 4.56
R superior temporal gyrus 42 68 ⫺30 14 4.23
R inferior temporal gyrus 37 62 ⫺54 ⫺4 3.96
R putamen // 30 ⫺12 ⫺6 3.95
R inferior frontal gyrus 47 44 36 ⫺2 3.67
L precentral gyrus 4 ⫺44 ⫺14 36 ⬍0.001 7.00 12,422
L superior temporal gyrus 22 ⫺48 ⫺22 4 6.79
L superior temporal gyrus 42 ⫺60 ⫺26 10 6.53
L precentral gyrus 6 ⫺40 ⫺18 66 5.64
L medial frontal gyrus 32 ⫺8 10 42 5.28
L postcentral gyrus 3 ⫺22 ⫺26 68 5.05
L insula 13 ⫺40 4 ⫺2 4.90
R medial frontal gyrus 32 8 10 44 4.82
R cingulate gyrus 24 ⫺4 0 46 4.74
L putamen // ⫺26 ⫺10 4 4.56
L inferior parietal lobe 40 ⫺50 ⫺32 58 4.52
R cerebellum // 16 ⫺58 ⫺24 ⬍0.001 5.77 1663
L cerebellum // ⫺46 ⫺62 ⫺32 4.48
R inferior parietal lobe 40 52 ⫺46 50 ⬍0.03 4.37 249
L fusiform gyrus 18 ⫺22 ⫺86 ⫺18 ⬍0.04 3.63 220
Regions more active for difficult rules than easy rules (in rule acquisition phases)
masked by the main effect of difficult rules minus easy rules
R superior parietal lobe 7 20 ⫺64 54 ⬍0.001 4.39 1199
L superior parietal lobe 7 ⫺18 ⫺64 54 3.93
R precuneus 7 30 ⫺72 50 3.92
R inferior parietal lobe 40 36 ⫺50 50 3.77
L precuneus 7 ⫺26 ⫺70 52 3.52
L inferior parietal lobe 40 ⫺40 ⫺56 46 3.46
L superior frontal gyrus 6 ⫺22 4 68 ⬍0.003 4.37 441
L middle frontal gyrus 9 ⫺46 26 34 ⬍0.001 3.98 541
L middle frontal gyrus 6 ⫺48 8 44 3.92
Stereotactic MNI coordinates for significant clusters (random effects, cluster-level p ⬍ 0.05 corrected, estimated at p ⬍ 0.001 uncorrected) given in millimeter with effect sizes (z scores) and cluster extent. In the Voxels per cluster column,
cluster extent is reported in correspondence of the main peak. Subpeaks were selected dividing each cluster into Brodmann areas (BA) and then selecting peaks within each area.

vs mid-DLPFC: F(4,76) ⫽ 26.91, p ⬍ 0.001; postMTG vs SPL: ported above was obtained when only the ROIs in the left hemi-
F(4,76) ⫽ 16.08, p ⬍ 0.001; and postMTG vs FPC: F(4,76) ⫽ 7.01, sphere were considered in the analyses.
p ⬍ 0.001). Critically, however, the region by condition interac-
tion was also significant for the mid-DLPFC versus FPC compar- Analysis of the effect of rule difficulty
ison (F(4,76) ⫽ 10.16; p ⬍ 0.001). There was also a trend for a Finally, in the whole-brain data, we investigated whether there
significant region by condition interaction between mid-DLPFC were brain regions modulated by rule difficulty. We first tested
and SPL (F(4,76) ⫽ 2.96; p ⫽ 0.025), but not for the FPC versus for the main effect of difficult rules minus easy rules, considering
SPL comparison (F(4,76) ⫽ 1.82; p ⫽ 0.13). Returning to the crit- all the four phases (from SearchPhase2 to FollowingPhase2) for
ical region by condition interaction for the mid-DLPFC versus which the two types of rule were analyzed separately. This con-
FPC comparison, post hoc tests (Bonferroni correction applied) trast revealed activation in several clusters (Table 4, first con-
showed that the two regions differed during rule following trials trast). The largest one had its peak of activation in the left
( p ⬍ 0.01 for both FollowingPhase1 and FollowingPhase2 trials) precentral gyrus and extended to the left superior temporal gyrus
but not during rule acquisition trials (all p ⬎ 0.6). Post hoc tests as well as to regions in the superior portion of the medial frontal
also showed a trend of FollowingPhase1 ⬎ FollowingPhase2 in gyrus. Similar regions were also activated in the right hemisphere
the FPC ( p ⫽ 0.07); this was in line with the same effect found in as part of a second large cluster. Activation in this cluster ex-
a similar region of the left hemisphere in the whole-brain cor- tended to the right inferior frontal gyrus. Moreover, both the
rected data. Finally, post hoc tests (Bonferroni correction applied) cerebellum and the lateral part of the inferior parietal lobe were
performed for each of the three two-factor interactions involving also more active (bilaterally) for difficult than easy rules.
the postMTG showed that this region was generally more active Of interest, some of the regions globally modulated by rule
than the other three regions specifically during rule following difficulty were similar to the regions contributing more to rule
phases, particularly during FollowingPhase2 trials. following than rule acquisition (i.e., superior temporal gyrus,
Finally, it should be noted that an analogous pattern of cingulate gyrus, cerebellum, putamen, and regions in the poste-
results for the two-factor region by condition interactions re- rior/lateral part of the precentral gyrus) (see above, Rule follow-
7772 • J. Neurosci., May 25, 2011 • 31(21):7763–7774 Crescentini et al. • Neural Markers of Rule Acquisition/Rule Following

analyses related to rule difficulty show a greater activation of

temporal and motor cortices for difficult than easy rules. More
generally, the data indicate that when a region is active, it is more
active for difficult than easy rules. Thus, we also found increased
activation of the left mid-DLPFC and of the left and right parietal
cortex for difficult rules during the acquisition phases. However,
this generalization does not apply to the right mid-DLPFC or the
medial/anterior PFC, which were activated to a similar extent by
easy and difficult rules during the rule acquisition and rule fol-
lowing phases, respectively (see also later in Discussion).

Discussion
In the present study, we investigated mechanisms of rule acqui-
sition and rule following in inductive reasoning. We used the
Brixton test (Burgess and Shallice, 1996) in which subjects attain
and then follow 30 different rules. Five distinct phases could be
clearly distinguished during performance on each successfully
attained rule. Whereas the first three phases tapped rule acqui-
sition, the remaining two involved rule following. Comparing
between the Acquisition part and the Following part (Search-
Phases1&2 and Discovery minus FollowingPhases1&2) revealed
Figure 5. Activity in mid-DLPFC, FPC, SPL, and postMTG ROIs for each phase identified in the
bilateral activation in frontal (mid-DLPFC and IFG/pre-PMd)
test and for fixation baseline blocks (Fix). Activity is measured by percentage signal change. The and parietal regions. These regions, except for the right mid-
vertical bars denote SD of the means. The location of each ROI is reported in top part of the figure DLPFC, were modulated by rule difficulty, being more active for
(the left hemisphere is shown). The signal was averaged across the right and left hemispheres in difficult than easy rules during rule acquisition. The opposite
each ROI. SrP1, SrP2, Dsv, FlP1, and FlP2 refer to SearchPhase1, SearchPhase2, Discovery, Fol- comparison (Following minus Acquisition) produced bilateral
lowingPhase1, and FollowingPhase2, respectively. activation in the temporal and precentral gyri, and in the medial/
anterior PFC. Orthogonal ROI analyses confirmed this pattern of
ing versus rule acquisition) (Tables 3, 4). It is important to note, results, showing in addition a dissociation between FPC and mid-
however, that an effect of rule following minus rule acquisition DLPFC. This was attributable to the prolonged activation of the
was found in these regions even when difficult and easy rules were FPC from rule acquisition trials to initial rule following trials (i.e.,
considered separately (contrasts not shown). In contrast, regions FollowingPhase1 trials). Moreover, the selective response of the
in the middle frontal gyrus (bilaterally) and in the left inferior postMTG during rule following dissociated this region from the
frontal gyrus, which showed increased activation for rule acqui- mid-DLPFC, FPC, and SPL ones.
sition relative to rule following in the basic analysis (Table 2), Activation of a network of frontoparietal regions in the Brix-
were not overall more active for difficult than easy rules. None- ton test is consistent with results of previous fMRI studies on
theless, the regions in the parietal cortex, especially in the right reasoning (Hampshire et al., 2011) and spatial working memory
hemisphere, which showed an effect of rule acquisition ⬎ rule (Smith and Jonides, 1997; D’Esposito et al., 1998), as well as with
following, were reasonably close to the right inferior parietal lobe the hypothesis that this network constitutes an adaptable system
region showing a general effect of rule difficulty. that is recruited in novel situations whenever the general level of
Next, we tested for the two simple main effects of difficult difficulty and the need for executive control increases (Hamp-
minus easy rules separately for the rule acquisition phases (i.e., shire et al., 2008). Nevertheless, our results extend those of pre-
SearchPhase2 and Discovery) and the rule following phases (i.e., vious studies by showing that frontal, parietal, and temporal
FollowingPhase1 and FollowingPhase2). We used the main effect regions differed in their patterns of activation during the five
contrast just described as an exclusive mask for both the two phases in which we divided the test.
contrasts. The first contrast concerned rule acquisition and The pattern of activation found in the mid-DLPFC fits with a
showed activation in three clusters (Table 4, second contrast). role for this region in rule acquisition rather than rule following.
One was centered in the right superior parietal lobe and involved During the Acquisition part of the Brixton test, participants tend
the same region in the left hemisphere as well as the inferior to generate at least one possible hypothesis for each card pre-
parietal lobe bilaterally. A second cluster involved the left supe- sented; on this view the mid-DLPFC appears to be critical for
rior frontal gyrus, whereas the third was centered in the left (but detecting regularities across stimuli and for using this informa-
not the right) middle frontal gyrus and involved the mid-DLPFC tion to generate appropriate hypotheses. The increased activation
and the IFJ/pre-PMd regions. Of importance, these latter regions of this area (in the left hemisphere) more for difficult than easy
as well as the parietal ones, were very similar to those showing a rules indicates that abstracting an organizing rule is particularly
main effect of rule acquisition minus rule following in the main demanding in difficult cases. A number of alternative hypotheses
analysis (see above, Rule acquisition versus rule following) (Ta- must often be generated before this type of rule can be correctly
bles 2, 4). The second simple main effect concerned the rule induced.
following phases and did not reveal any significant activation. This generalization is in line with previous research (Reverberi
Finally, it should be noted that the FPC appeared not to be et al., 2005a,b; Specht et al., 2009), but differs in emphasis from
modulated by rule difficulty overall. However, this region is par- conclusions from other studies, including those using neuro-
ticular in not showing an effect of rule acquisition minus rule physiological recordings in nonhuman primates. Thus, although
following in the main analysis. Thus, it is insensitive to the con- some studies (Wang et al., 2000; Bussey et al., 2001) have shown
trasts examined in the previous paragraph. In sum, the additional that lesioning ventral PFC after learning does not affect subse-
Crescentini et al. • Neural Markers of Rule Acquisition/Rule Following J. Neurosci., May 25, 2011 • 31(21):7763–7774 • 7773

quent rule use in monkeys, other studies, in which recording was in the Brixton test when one needs to represent the outcomes
from regions in the principal sulcus extending to DLPFC (BA of multiple separate inferences simultaneously (Ramnani and
9/46), have supported the idea that this latter region controls the Owen, 2004; Bunge et al., 2009).
guidance of behavior according to well learned rules (White and We now turn to the fMRI analysis that dealt with activity
Wise, 1999; Asaad et al., 2000; Mansouri et al., 2006). From this during rule following minus rule acquisition. Here, we found
perspective, mid-DLPFC, as well as more posterior brain regions activation in regions such as temporal and motor cortex that
such as the IFJ and PMd, would mediate rule implementation by appear as likely candidates for rule storage and motor planning/
selecting and maintaining the appropriate response contingen- execution, respectively (Bunge, 2004; Donohue et al., 2005). The
cies associated with the currently relevant task rules (Bunge, activation found in the medial/anterior PFC may reflect, how-
2004; Brass et al., 2005; Bunge et al., 2005). ever, default-mode network (DMN) activity (Raichle, 2010). A
An alternative possibility, however, that cannot be ruled out is linear decrease in activation of the midline DMN regions has
that the mid-DLPFC could have played a general monitoring role been observed with increasing task difficulty and working mem-
in our task. Monitoring would be required during the acquisition ory demands (Esposito et al., 2006; Sambataro et al., 2010). Our
phases to overcome the tendency to repeat inappropriate rules/ study supports the position that the deactivation observed in this
hypotheses (Reverberi et al., 2005a). region is task induced; there is a linear decrease in activation from
The pattern of activation that we found for the FPC indicates SearchPhase1 up to Discovery, a period over which the number
that, unlike the mid-DLPFC, this region contributes to inductive of instances and hypotheses grows; there is also a correlation
reasoning after hypothesis generation. We suggest that, in this between activation in this region and RT during Discovery trials.
task, FPC maintained alternative hypotheses in a pending state Similarly, the medial/anterior PFC is the one region (together
during rule acquisition, until the positive feedback received after with the right mid-DLPFC) that breaks the following rule: if a
the Discovery and FollowingPhase1 phases, unambiguously in- region is positively activated in a phase, then its activation is in-
dicates the validity of one hypothesis. Take, for instance, the cen- creased for the more difficult rules. This does not apply to the medi-
tral square rule reported in Table 1 (rule 25); a typical situation al/anterior PFC in FollowingPhase1 and FollowingPhase2 trials. For
for the first correct response is when the blue circle is predicted to other regions, greater relevant resource requirements (e.g., hypoth-
be in position 10. On this card, the participant should consider at esis generation, rule storage) lead to greater activations.
least two alternative rules producing number 9 (correct) and then In conclusion, the present study provides evidence that the
number 3, in the first case, and numbers 11 and 5, in the second. network of frontal, parietal, and temporal regions underlying
The FPC would still be activated when the blue circle is in posi- inductive reasoning fractionates with different subregions re-
tion 9 (FollowingPhase1 trial). sponding to rule acquisition and rule following phases. Mid-
The present account for the FPC activation is in line with DLPFC was consistently activated during rule acquisition; FPC
other proposals concerning the region. Thus, Burgess et al. (2000, contributed to rule acquisition but also to the initial phases of
2001) argued that the FPC is critical for the setting up and real- rule following. Parietal cortex showed a pattern similar to FPC.
ization of intentions. Boorman et al. (2009) recently strength- By contrast, motor and temporal cortex contributed particularly
ened this idea by showing that this region is important in to rule following.
representing a pending or alternative potential course of action to
determine when to switch to that behavior. On this account, the References
parietal lobes would then be responsible for implementing the Asaad WF, Rainer G, Miller EK (2000) Task-specific neural activity in the
switches. It is striking that, in our study, the parietal lobes too primate prefrontal cortex. J Neurophysiol 84:451– 459.
remained relatively active during FollowingPhase1 trials. Badre D, D’Esposito M (2007) Functional magnetic resonance imaging ev-
Recently, there has been a growing consensus that PFC may be idence for a hierarchical organization of the prefrontal cortex. J Cogn
Neurosci 19:2082–2099.
functionally organized to support progressively more abstract Badre D, Kayser AS, D’Esposito M (2010) Frontal cortex and the discovery
representations along a rostrocaudal axis from PMd to FPC cor- of abstract action rules. Neuron 66:315–326.
tex, passing through pre-PMd and mid-DLPFC (Badre and Boettiger CA, D’Esposito M (2005) Frontal networks for learning and execut-
D’Esposito, 2007). Overall, our findings of sustained activation of ing arbitrary stimulus-response associations. J Neurosci 25:2723–2732.
the pre-PMd and mid-DLPFC during rule acquisition and the Boorman ED, Behrens TE, Woolrich MW, Rushworth MF (2009) How
protracted activation of the FPC during FollowingPhase1 trials green is the grass on the other side? Frontopolar cortex and the evidence in
favor of alternative courses of action. Neuron 62:733–743.
are in agreement with the proposed hierarchical organization of
Brainard DH (1997) The Psychophysics Toolbox. Spat Vis 10:433– 436.
the PFC. Brass M, Derrfuss J, Forstmann B, von Cramon DY (2005) The role of the
In this context, in a recent study, Badre et al. (2010) studied inferior frontal junction area in cognitive control. Trends Cogn Sci
the neural mechanisms of abstract rule learning and again found 9:314 –316.
a rostrocaudal pattern of activation in the PFC. However, unlike Brett M, Anton JL, Valabregue R, Poline JB (2002) Region of interest anal-
our study, in that by Badre et al., abstract rule learning was me- ysis using an SPM toolbox. Paper presented at Eighth International Con-
ference on Functional Mapping of the Human Brain, Sendai, Japan, June.
diated by processing in the pre-PMd region and activity in the
Bunge SA (2004) How we use rules to select actions: a review of evidence
PMd (and in the mid-DLPFC and FPC regions) did not distin- from cognitive neuroscience. Cogn Affect Behav Neurosci 4:564 –579.
guish between the hierarchical (abstract) rule set and the flat rule Bunge SA, Wallis JD (2008) Neuroscience of rule-guided behavior. Oxford:
set they used. It is possible that the difference between our study Oxford UP.
and that by Badre et al. in the pattern of results reflects differences Bunge SA, Wallis JD, Parker A, Brass M, Crone EA, Hoshi E, Sakai K (2005)
in the type of rules and how they are learned, namely by inductive Neural circuitry underlying rule use in humans and nonhuman primates.
reasoning and reinforcement learning, respectively. From this J Neurosci 25:10347–10350.
Bunge SA, Helskog EH, Wendelken C (2009) Left, but not right, rostrolat-
perspective, rules in the Brixton test would be sufficiently abstract eral prefrontal cortex meets a stringent test of the relational integration
to require the involvement of the mid-DLPFC and the FPC cor- hypothesis. Neuroimage 46:338 –342.
tex. On this account, the latter area would be at the apex of our Burgess PW, Shallice T (1996) Bizarre responses, rule detection and frontal
ability to process abstract representations such as those involved lobe lesions. Cortex 32:241–259.
7774 • J. Neurosci., May 25, 2011 • 31(21):7763–7774 Crescentini et al. • Neural Markers of Rule Acquisition/Rule Following

Burgess PW, Veitch E, de Lacy Costello A, Shallice T (2000) The cognitive Nichols T, Brett M, Andersson J, Wager T, Poline JB (2005) Valid conjunc-
and neuroanatomical correlates of multitasking. Neuropsychologia tion inference with the minimum statistic. Neuroimage 25:653– 660.
38:848 – 863. Raichle ME (2010) Two views of brain function. Trends Cogn Sci 14:180 –190.
Burgess PW, Quayle A, Frith CD (2001) Brain regions involved in prospec- Ramnani N, Owen AM (2004) Anterior prefrontal cortex: insights into
tive memory as determined by positron emission tomography. Neuropsy- function from anatomy and neuroimaging. Nat Rev Neurosci 5:184 –194.
chologia 39:545–555. Reverberi C, Lavaroni A, Gigli GL, Skrap M, Shallice T (2005a) Specific
Bussey TJ, Wise SP, Murray EA (2001) The role of ventral and orbital pre- impairments of rule induction in different frontal lobe subgroups. Neu-
frontal cortex in conditional visuomotor learning and strategy use in ropsychologia 43:460 – 472.
rhesus monkeys (Macaca mulatta). Behav Neurosci 115:971–982. Reverberi C, D’Agostini S, Skrap M, Shallice T (2005b) Generation and rec-
D’Esposito M, Aguirre GK, Zarahn E, Ballard D, Shin RK, Lease J (1998) ognition of abstract rules in different frontal lobe subgroups. Neuropsy-
Functional MRI studies of spatial and nonspatial working memory. Brain chologia 43:1924 –1937.
Res Cogn Brain Res 7:1–13. Rips LJ (1999) Deductive reasoning. In: The MIT encyclopedia of the cog-
Donohue SE, Wendelken C, Crone EA, Bunge SA (2005) Retrieving rules nitive sciences (Wilson RA, Keil FC, eds), pp 225–226. Cambridge, MA:
for behavior from long-term memory. Neuroimage 26:1140 –1149. MIT.
Esposito F, Bertolino A, Scarabino T, Latorre V, Blasi G, Popolizio T, Tedeschi Rissanen J (1978) Modeling by shortest data description. Automatica 14:
G, Cirillo S, Goebel R, Di Salle F (2006) Independent component model 465– 471.
Sambataro F, Murty VP, Callicott JH, Tan HY, Das S, Weinberger DR, Mattay
of the default-mode brain function: assessing the impact of active think-
VS (2010) Age-related alterations in default mode network: impact on
ing. Brain Res Bull 70:263–269.
working memory performance. Neurobiol Aging 31:839 – 852.
Friston KJ, Glaser DE, Henson RN, Kiebel S, Phillips C, Ashburner J (2002)
Seger CA, Poldrack RA, Prabhakaran V, Zhao M, Glover GH, Gabrieli JD
Classical and Bayesian inference in neuroimaging: applications. Neuro-
(2000) Hemispheric asymmetries and individual differences in visual
image 16:484 –512.
concept learning as measured by functional MRI. Neuropsychologia
Goel V, Dolan RJ (2000) Anatomical segregation of component processes in
38:1316 –1324.
an inductive inference task. J Cogn Neurosci 12:110 –119.
Smith EE, Jonides J (1997) Working memory: a view from neuroimaging.
Hampshire A, Thompson R, Duncan J, Owen AM (2008) The target selec- Cogn Psychol 33:5– 42.
tive neural response—similarity, ambiguity, and learning effects. PLoS Solomonoff R (1964) A formal theory of inductive inference, part I. Inform
One 3:e2520. Control 7:1–22.
Hampshire A, Thompson R, Duncan J, Owen AM (2011) Lateral prefrontal Specht K, Lie CH, Shah NJ, Fink GR (2009) Disentangling the prefrontal
cortex subregions make dissociable contributions during fluid reasoning. network for rule selection by means of a non-verbal variant of the Wis-
Cereb Cortex 21:1–10. consin Card Sorting Test. Hum Brain Mapp 30:1734 –1743.
Henson R (2005) What can functional neuroimaging tell the experimental Strange BA, Henson RN, Friston KJ, Dolan RJ (2001) Anterior prefrontal
psychologist? Q J Exp Psychol A 58:193–233. cortex mediates rule learning in humans. Cereb Cortex 11:1040 –1046.
Koechlin E, Hyafil A (2007) Anterior prefrontal function and the limits of Wang M, Zhang H, Li BM (2000) Deficit in conditional visuomotor learn-
human decision-making. Science 318:594 –598. ing by local infusion of bicuculline into the ventral prefrontal cortex in
Koechlin E, Basso G, Pietrini P, Panzer S, Grafman J (1999) The role of the monkeys. Eur J Neurosci 12:3787–3796.
anterior prefrontal cortex in human cognition. Nature 399:148 –151. White IM, Wise SP (1999) Rule-dependent neuronal activity in the prefron-
Li M, Vitányi P (1997) An introduction to Kolmogorov complexity and its tal cortex. Exp Brain Res 126:315–335.
applications. New York: Springer. Zaitsev M, Hennig J, Speck O (2004) Point spread function mapping with
Mansouri FA, Matsumoto K, Tanaka K (2006) Prefrontal cell activities re- parallel imaging techniques and high acceleration factors: fast, robust,
lated to monkeys’ success and failure in adapting to rule changes in a and flexible method for echo-planar imaging distortion correction. Magn
Wisconsin Card Sorting Test analog. J Neurosci 26:2745–2756. Reson Med 52:1156 –1166.