doi:10.1093/brain/awl017
Brain (2006), 129, 944–952
Deciding how to decide: ventromedial frontal
lobe damage affects information acquisition in
multi-attribute decision making
Lesley K. Fellows
Montreal Neurological Institute, rue Université, Montréal, QC, Canada
Correspondence to: Lesley K. Fellows, Montreal Neurological Institute, Room 276, 3801 rue Université, Montréal,
QC H3A 2B4, Canada
E-mail: [email protected]
Ventromedial frontal lobe (VMF) damage is associated with impaired decision making. Recent efforts to understand the functions of this brain region have focused on its role in tracking reward, punishment and risk.
However, decision making is complex, and frontal lobe damage might be expected to affect it at other levels.
This study used process-tracing techniques to explore the effect of VMF damage on multi-attribute decision
making under certainty. Thirteen subjects with focal VMF damage were compared with 11 subjects with frontal
damage that spared the VMF and 21 demographically matched healthy control subjects. Participants chose
rental apartments in a standard information board task drawn from the literature on normal decision making.
VMF subjects performed the decision making task in a way that differed markedly from all other groups,
favouring an ‘alternative-based’ information acquisition strategy (i.e. they organized their information search
around individual apartments). In contrast, both healthy control subjects and subjects with damage predominantly involving dorsal and/or lateral prefrontal cortex pursued primarily ‘attribute-based’ search strategies
(in which information was acquired about categories such as rent and noise level across several apartments).
This difference in the pattern of information acquisition argues for systematic differences in the underlying
decision heuristics and strategies employed by subjects with VMF damage, which in turn may affect the quality
of their choices. These findings suggest that the processes supported by ventral and medial prefrontal cortex
need to be conceptualized more broadly, to account for changes in decision making under conditions of
certainty, as well as uncertainty, following damage to these areas.
Keywords: executive function; problem solving; prefrontal cortex; human, lesion
Abbreviations: D-CTL = dorsal/lateral control group; D/LF = dorsal and/or lateral frontal lobe;
V-CTL = ventromedial control group; VMF = ventromedial frontal lobe
Received September 28, 2005. Revised December 23, 2005. Accepted January 2, 2006. Advance Access publication February 2, 2006.
Introduction
Decision making, the process of choosing between options,
is a fundamental human behaviour. In the last several years,
cognitive neuroscience studies have begun to investigate the
brain basis of decision making, with a particular focus on
prefrontal cortex. This focus has been motivated, in part,
by clinical reports that frontal damage [and particularly
ventromedial frontal (VMF) damage] can be associated with
strikingly poor decision making (Eslinger and Damasio,
1985; Harlow, 1999; Ackerly, 2000). The most influential
laboratory studies of decision making after VMF damage
have focused on risky decisions under uncertainty, as captured by a card-based gambling task (Bechara et al., 1994,
#
1997). Although this task does detect abnormal behaviour
in subjects with VMF damage, there is ongoing controversy
about these findings and their interpretation (Manes et al.,
2002; Tomb et al., 2002; Sanfey et al., 2003; Maia and
McClelland, 2004, 2005; Bechara et al., 2005; Fellows and
Farah, 2005).
However, in some respects this controversy bypasses a
larger issue. Decision making is a complex behaviour that
clearly involves multiple component processes (Krawczyk,
2002; Fellows, 2004; Sugrue et al., 2005). In principle, the
poor real-life decisions of VMF-damaged subjects could be
the result of deficits in any one or more than one of these
The Author (2006). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: [email protected]
Multi-attribute decision making following frontal damage
processes; there is no a priori reason to expect that the
impaired decision making of such individuals rests solely,
or even primarily, on a specific deficit in dealing with risk
or uncertainty. It is particularly important to clarify the
scope of decision-making impairments following frontal
lobe damage, because such evidence can be very helpful in
interpreting the data being generated in the burgeoning functional imaging and electrophysiology literatures examining
the role of ventral and medial prefrontal cortex in various
aspects of decision making (Critchley et al., 2001; Ernst et al.,
2002; Gehring and Willoughby, 2002; Krawczyk, 2002;
McClure et al., 2004a; Tanaka et al., 2004). The present
study therefore represents an effort to move away from a
narrow focus on risky decision making, and instead examine
the effects of focal frontal lobe damage on a form of decision
making under certainty: multi-attribute decision making.
Decision making does not have to involve risk to be hard.
Anyone who has purchased a car, chosen a school to attend
or hired an employee from amongst several applicants knows
that weighing the pros and cons of options that differ across
multiple, sometimes incommensurate, dimensions can be very
difficult, even when all the information is known and the
potential outcomes are certain. This kind of decision making
has been examined in normal subjects, including studies of
the information acquisition strategies such subjects use, the
determinants of their choice of strategy and the quality of the
decisions that result (Simon, 1955; Payne, 1976; Ford et al.,
1989; Payne et al., 1993; Chu et al., 1999; Gigerenzer and
Todd, 1999). This work has shown that individuals may
use a variety of strategies and heuristics to cope with the
potential complexity of such decisions (Payne et al., 1992,
1993; Gigerenzer and Todd, 1999). Indeed, in such settings
‘deciding how to decide’ can become the central problem.
How many alternatives should be considered? How much
information should be learned about each? Should one
search for the best alternative or settle for one that is ‘good
enough’? Importantly, different solutions to these predecisional problems often result in different final choices
(Payne et al., 1992), and in different levels of satisfaction
regarding these choices (Schwartz, 2004).
One view emerging from process-oriented studies of
decision making is that these strategic pre-decisional processes are flexible and adaptive, often constructed ‘on the fly’
as subjects engage in information acquisition. External constraints, such as instructions, information complexity and
time pressure, can influence such strategies. In addition, internal factors such as processing capacity and individual differences in decision goals (which may themselves be flexibly
determined and updated as more information is acquired)
also contribute (Ford et al., 1989; Payne et al., 1992;
Gigerenzer and Todd, 1999; Simon et al., 2004).
Thus, multi-attribute decision making involves a dynamic
interplay between assessing the current relative value of
alternatives, efficiently acquiring further information to optimize this assessment of value and adjusting overall decision
goals in light of the available information. As such, this form
Brain (2006), 129, 944–952
945
of decision making would seem to call upon several capacities
linked to prefrontal cortex, even in the absence of risk or
uncertainty. Further, the study of this aspect of decision making may constitute a link between the literature examining
prefrontal contributions to planning and problem solving
in open-ended settings (Shallice and Burgess, 1991; Goel
et al., 1997; Goel and Grafman, 2000) and recent work on
the role of the VMF region in representing the relative value of
stimuli in simple associative learning paradigms (reviewed in
Rolls, 2000; Fellows, 2004; Sugrue et al., 2005).
This study used a simple process-tracing methodology
adapted from studies of decision making in normal subjects
to examine multi-attribute decision making in individuals
with focal frontal lobe damage. The performance of those
with VMF damage was contrasted with a lesioned control
group with frontal damage that spared the VMF, and with
healthy age-matched control subjects. The author asked
whether damage to either frontal region is associated
with systematic differences in the strategic, informationacquisition phase of such decision making.
Methods
Subjects
The study involved 13 subjects with damage to the VMF and
11 subjects with damage to dorsal and/or lateral frontal lobe (D/LF)
(Fig. 1). These two groups differed in educational level, and so were
compared to separate demographically matched healthy control
groups [ventromedial control group (V-CTL), n = 11; dorsal/lateral
control group (D-CTL), n = 10]. Subjects with frontal damage
were identified through the University of Pennsylvania Center for
Cognitive Neuroscience patient database. VMF damage was due to
rupture of anterior communicating aneurysm in 10 cases, and to
ischaemic stroke in 3. D/LF damage followed ischaemic or haemorrhagic stroke in eight cases, and resection of low-grade glioma with
local radiotherapy in three. Five VMF subjects and five D/LF subjects were taking psychoactive medications. These were most commonly anticonvulsants and/or antidepressants. One VMF subject
was taking an acetylcholinesterase inhibitor, another both an
acetylcholinesterase inhibitor and methylphenidate, and a third
modafenil. Subjects were tested at least 6 months (mean, 4.2 years,
range 0.5–12 years) after brain injury had occurred.
Lesions were traced from the most recent CT or MR imaging
available onto the standard Montreal Neurological Institute brain
using MRIcro software (Rorden and Brett, 2000). Clinical imaging
was used for this purpose and so pre-dated participation in this
study by months to years, depending on the recency of the brain
injury, and the intensity of clinical follow-up of individual patients.
At the time of study participation, no subjects had clinical evidence
(on history or neurological examination) of central nervous
system injury beyond that associated with the index lesion.
Subjects were assigned a priori to the VMF group if the lesion
principally affected medial orbitofrontal and/or the ventral aspect
of the medial prefrontal cortex [following the boundaries laid out in
Stuss and Levine (2002)]. Those with a lesion involving the frontal
lobe anterior to the pre-central sulcus, but sparing VMF areas, were
assigned to the D/LF group. As can be seen in Fig. 1, those in the D/LF
group most commonly had damage affecting the inferior and
946
Brain (2006), 129, 944–952
L. K. Fellows
Fig. 1 Location and overlap of brain lesions. The top row shows the lesions of the 13 subjects with ventromedial frontal damage; the
bottom row shows those of the 11 D/LF subjects. Lesions are projected on the same seven axial slices of the standard Montreal
Neurological Institute brain for both groups, oriented according to radiological convention (right is left). Areas damaged in one subject are
shown in purple; warmer shades denote the degree to which lesions involve the same structures in two or more individuals, as indicated in
the key. Areas of damage common to both groups can be assessed by comparing each slice to the slice directly beneath it, and are described
in more detail in the text.
Table 1 Subject characteristics; see text for details [mean (SD)]
Group
Age
(years)
Education
(years)
Estimated IQ
Beck
Depression
Inventory
Frontal lesion
volume (cc)
VMF (n = 13)
V-CTL (n = 11)
D/LF (n = 11)
D-CTL (n = 10)
56.1
60.4
58.3
57.1
13.0
13.1
15.6
16.4
115
119
121
128
12.4
6.0
9.9
3.6
22 (20)
(11.4)
(8.9)
(11.2)
(12.7)
(2.2)
(1.2)
(2.6)
(1.8)
middle frontal gyrus, and none had damage affecting either medial
orbitofrontal cortex or that portion of medial prefrontal cortex
ventral to the genu of the corpus callosum. Two VMF subjects
had damage that extended into D/LF areas: one with involvement
of the lateral aspect of the right inferior frontal gyrus and another
with extensive bilateral frontopolar and dorsomedial prefrontal
damage in addition to near-complete destruction of orbitofrontal
and ventral medial prefrontal cortex.
Healthy control subjects were recruited by advertisement. Controls were not taking psychoactive medication and were free of
significant current or past psychiatric or neurological illness as
determined by history and screening neurological examination.
Controls were excluded if they scored less than 28 out of 30 on
the mini-mental status examination (Folstein et al., 1983). IQ was
estimated by means of the American version of the National Adult
Reading Test. All subjects provided written, informed consent prior
to participation in the study, in accordance with the Declaration of
Helsinki, and were paid a nominal fee for their time. The Institutional Review Board of the University of Pennsylvania approved the
study protocol.
Demographic information is summarized in Table 1. The frontal
groups did not differ significantly from their respective control
groups in age, education or estimated IQ (t-test, all P > 0.05).
The D/LF group had significantly more education than the VMF
group (P < 0.05), but did not differ significantly in estimated IQ or in
(8)
(10)
(11)
(6)
(8.7)
(2.3)
(4.7)
(4.0)
36 (36)
volume of damaged tissue (all P > 0.05). The D/LF group scored
significantly higher on the Beck Depression Inventory than the
D-CTL group (P < 0.01), as did the VMF compared with the
V-CTL group (P < 0.05), but the two frontal groups did not differ
significantly on this measure (P > 0.05).
Subjects with frontal damage were administered a short neuropsychological battery for screening purposes. Results from the tasks
with potential sensitivity to frontal damage, as well as a verbal
memory task (recall of a list of five words after a 1 min delay),
are provided in Table 2 (not all subjects completed all tests). The
groups differed significantly only in their performance on the
Trails B task, with VMF subjects making more errors (Mann–
Whitney U-test, P < 0.01). The control groups also completed the
animal and ‘F’ verbal fluency tasks. The D/LF group was significantly
impaired on both fluency tasks compared with the D-CTL group
(P < 0.01), whereas the VMF group did not differ from the V-CTL
group on either measure.
Decision task
The pattern of information search in multi-attribute decision
making was assessed using a standard ‘information board’ paradigm
(Payne, 1976). Subjects made hypothetical choices between onebedroom apartments. Information about the apartments was displayed in table format on a computer screen. Initially, only the
Multi-attribute decision making following frontal damage
Brain (2006), 129, 944–952
947
Table 2 Results of selected neuropsychological screening tests [mean (SD)]
Group
Digit span forward
Animal fluency
‘F’ fluency
Trails B errors
Verbal recall
VMF
V-CTL
D/LF
D-CTL
5.0 (0)
15.4
19.6
16.1
24.1
10.7
11.9
10.9
17.1
3.6 (2.8)
3.4 (1.5)
0.8 (0.9)
3.6 (1.0)
5.2 (1)
(5.4)
(7.3)
(4.6)
(4.6)
(5.4)
(3.6)
(3.9)
(5.6)
down)/(movements across + movements down). This results in a
score that ranges from 1 to +1, capturing the degree to which the
board was searched ‘down’ (by alternatives), or ‘across’ (by attributes), regardless of how much information was acquired. An
entirely alternative (i.e. apartment)-based search strategy would
yield a score of 1, whereas an entirely attribute-based strategy
would yield a score of +1. The percentage of the available information
examined and the final choice were also recorded.
Statistical analysis
Fig. 2 Example of information board layout, showing a partial view
of the 4 option-6 attribute (4 · 6) task, with some of the
information revealed.
apartment labels (e.g. ‘apartment A’, as column headings) and the
attributes for which information was available (e.g. ‘neighborhood’,
as row headings) were shown (Fig. 2). The information itself was
masked by white rectangles. Subjects indicated which information
they wanted by clicking the mouse on the relevant white rectangle.
The rectangle was replaced by the information, which then remained
visible until a final decision was made, to eliminate any memory
requirement. Subjects were instructed to seek as much information
as they felt they needed, and in the order that made sense to them, to
make a decision as efficiently as possible, proceeding ‘just as they
would in real life’. They were further told that it was not necessary to
examine all the information and that they could decide on an apartment whenever they felt ready. They were also told that there was no
correct answer and that they should choose the apartment that was
right for them. There was no time limit. In cases where subjects were
uncomfortable using a mouse, they instead pointed to the rectangle
they wished to open, and the experimenter moved the mouse for
them. Subjects were also encouraged to ‘think aloud’, and their
comments were tape recorded, in keeping with the original
method of Payne (1976). However, participants with frontal lobe
damage made few or no informative comments, so these data were
not analysed further.
All subjects first performed a practice decision problem, choosing
between two cars, provided with information about two attributes.
They then selected an apartment from information boards of
increasing complexity: 2 apartments-4 attributes (2 · 4), 4 apartments-6 attributes (4 · 6), and 6 apartments-7 attributes (6 · 7). The
tasks were administered in the same order for all subjects. As is
typical for such tasks, there was no obviously best (or worst) choice.
The pattern of information acquisition was the main dependent
variable. This was summarized using the scoring system of Payne
(1976), according to the formula: (movements across movements
Subjects with frontal lobe damage were classified a priori into two
groups: VMF and D/LF. These two groups were similar in age,
but differed significantly in level of education, necessitating two
healthy control groups matched with the respective frontal-damaged
group on these two variables. Comparison of each frontal group with
the appropriate control group was carried out using t-tests where
the data conformed to a normal distribution, and Mann–Whitney
U-tests where they did not. Significance was set at P < 0.05,
two-tailed.
Results
Information search patterns
Subjects with VMF damage acquired information in a very
different way than did either those with D/LF damage or
either healthy control group. As shown in Fig. 3, VMF subjects predominantly pursued an alternative-based information search strategy, acquiring information regarding
multiple attributes of one alternative (e.g. apartment A),
then moving to another alternative. In contrast, the other
three groups of subjects pursued a primarily attributebased strategy, comparing several alternatives across one
attribute (e.g. rent), then moving to another attribute. The
total score of VMF subjects differed significantly from that
of the V-CTL group [mean score (SD) summed across all
three tasks: VMF 1.3 (2.3), V-CTL 2.1 (1.2), Mann–
Whitney U = 21, P < 0.01], whereas the scores of D/LF
and D-CTL groups did not (U = 49.5, P = 0.7). The same
pattern was evident in all three boards when analysed individually (Fig. 3).
Despite the differences in search strategies, all four
groups accessed a similar proportion of the available information overall (Table 3). In keeping with the literature
(Payne, 1976; Kerstholt, 1992), as the information content
of the decision boards increased, participants acquired
more information in absolute terms. However, they became
relatively more selective, examining a progressively smaller
948
Brain (2006), 129, 944–952
L. K. Fellows
proportion of the available information. This pattern did not
differ across groups [ANOVA (analysis of variance),
effect of group F(3,41) = 1.7, P > 0.05, effect of task complexity
F(2,42) = 30.6, P < 0.0001, task complexity · group, F(6,82) =
1.2, P > 0.05)]. In contrast, the pattern of information acquisition (search strategy) was not systematically affected by
increasing complexity.
Although there have been both anecdotal (Eslinger and
Damasio, 1985) and experimental (Rogers et al., 1999) reports
that VMF damage can lead to prolonged decision times,
this was not evident in the present experiment: The time
to complete each task increased with task complexity
(ANOVA, F(2,42) = 48.5, P < 0.0001), but there was no
significant effect of group, nor interaction (all P > 0.5;
Table 3).
The decision boards have no obvious right or wrong
answers and therefore provide no absolute measure of
choice quality, as is typical for such process-oriented tasks.
Nevertheless, some apartments were more popular than
Fig. 3 Median decision board scores for each group. Individual
bars show the score for each decision board problem, as
indicated by the key (above). More positive scores indicate a more
‘attribute-based’ search strategy, whereas more negative scores
indicate an ‘alternative-based’ search strategy.
others amongst the control subjects. Seventeen out of 21
control subjects chose apartment B in the first problem, 12
out of 21 chose apartment B in the second problem, and 11
out of 21 chose apartment C in the third problem (Table 3).
These were also the modal choices for the D/LF group. In
contrast, the VMF group tended to choose apartments that
were not the modal choice overall more often than the other
groups. This tendency was only statistically significant in the 4
· 6 task, where fewer VMF subjects chose the option that was
the modal choice of the group as a whole than did the V-CTL
subjects (Fisher’s exact test, P < 0.05). This suggests that the
markedly different search strategy of the VMF group influenced decision outcomes, as has been reported in normal
subjects (Payne et al., 1992).
Thus, subjects with VMF damage acquired information in
a systematically different way, and tended to make different
final choices. Although there was no statistically significant
difference in the amount of information gathered across
all four groups, there was considerable variability in this
measure. Clinical lore [and some systematic experimental
work (e.g. Berlin et al., 2004)] indicates that VMF damage
can lead to heightened impulsivity. Can the choices of those
with VMF damage in the present experiment be considered
‘impulsive’, in some sense of this variously defined term
(Evenden, 1999)? The lack of any group-wise differences in
the amount of information examined or in the time taken to
complete the tasks argues against this possibility. However,
individual patterns of behaviour in the simplest decision task
provide some support for this view: in the 2 · 4 task, in which
subjects chose between just 2 apartments, all control and all
D/LF subjects accessed at least some information about both
alternatives, with 29 out of 32 examining all of the eight pieces
of information available. Strikingly, 6 of 13 VMF subjects did
not examine all the information, and 4 of 13 only examined
information about one of the alternatives before choosing it.
There were no evident differences in lesion characteristics or
demographic variables distinguishing these subjects from the
VMF group as a whole. This behaviour could be interpreted as
impulsive, in the sense that subjects are making a decision
based on information that is incomplete compared with that
considered by the control group. An alternative explanation is
that VMF subjects had different decision goals: they were not
attempting to determine the relative merits of the different
alternatives, but were instead comparing a given alternative to
Table 3 Amount of information acquired and number of attributes examined summed across all three decision boards and
expressed as a percentage [mean (SD)].
Group
VMF
V-CTL
D/LF
D-CTL
Information (%)
71
84
66
76
(24)
(19)
(22)
(18)
Attributes (%)
93
95
89
94
(10)
(10)
(14)
(10)
Decision time (s)
Modal choice
2·4
4·6
6·7
2·4
4·6
6·7
65
66
71
58
113
107
152
110
174
146
177
138
B
B
B
B
A
B
B
B
C/E
C
C
C
(25)
(31)
(48)
(25)
(45)
(67)
(88)
(40)
(94)
(72)
(112)
(56)
The modal choice in each decision board for each group is shown in the third column. Groups did not differ significantly in the amount
of information, nor in the number of attributes they examined.
Multi-attribute decision making following frontal damage
some absolute standard of acceptability, an often effective
decision-making strategy known as ‘satisficing’ (Schwartz
et al., 2002).
The majority of the VMF group had at least some degree of
bi-hemispheric damage, precluding analysis of laterality
effects. The D/LF group included five individuals with left
hemisphere damage and six with right hemisphere damage.
There did not appear to be any systematic difference in search
strategy or amount of information examined in these two
subgroups. [Median total decision board score left D/LF
+2.1, right D/LF +1.8; mean (SD) proportion of information
examined left D/LF 0.67 (0.22), right D/LF 0.66 (0.24)].
Within the D/LF and VMF groups, lesion aetiology did not
appear to reliably predict search strategy, although such patterns would be difficult to detect given the small sample size.
Discussion
This was an exploratory study of the effects of frontal lobe
damage on the processes underlying multi-attribute decision
making. The intent was to more comprehensively delineate
the stages of decision making that are affected by such
damage. In contrast to other paradigms used in recent studies
of decision making in patients with frontal damage, multiattribute decision making does not involve risk, uncertainty
or learning.
Subjects with VMF damage successfully performed the
decision tasks: they acquired a similar amount of information
as the other groups and made decisions within a comparable
period of time (although their choices were somewhat different than the norm). However, the VMF group differed strikingly from all other groups in how they acquired information.
Importantly, this difference was systematic. Such patients did
not acquire information in a disorganized way, but rather
consistently applied an information acquisition strategy
that differed from that of other participants.
Process-oriented studies of multi-attribute decision making in normal subjects have argued that it is possible to draw
inferences about a subject’s underlying decision strategy by
measuring how much information is acquired, and the pattern of this information search (e.g. Payne et al., 1993). VMF
subjects acquired information following an alternative-based
(sometimes called inter-dimensional) pattern, in contrast to
the largely attribute-based (intra-dimensional) pattern followed by all other groups in the context of this task. That
is, VMF subjects tended to organize their information acquisition around individual apartments, whereas other groups
compared information about specific attributes, such as rent,
across several apartments. This finding suggests that VMF
damage leads to systematic differences in the strategies and
heuristics called upon in this form of decision making under
certainty.
How might VMF damage result in such strategy
differences? Two possibilities are suggested by separate
lines of evidence: first, that such damage leads to difficulties
Brain (2006), 129, 944–952
949
in managing information in minimally structured environments, similar to the impairments seen in ill-defined problem-solving tasks following frontal injury. Second,
(perhaps relatedly) that it results in an impaired ability to
determine the relative value of alternatives.
There is clearly an overlap between the ‘pre-decisional’,
information acquisition phase of multi-attribute decision
making and some forms of problem solving, both conceptually and in their experimental instantiations. Indeed, the
process-oriented approach to studying decision making used
here was originally adapted from studies of problem solving
(Newell and Simon, 1972). Broadly similar methods have
previously been used to study the effects of frontal lobe
damage on ill-structured problem solving. One such study
relied on a detailed analysis of ‘think aloud’ data generated
while subjects performed a hypothetical financial planning
task (Goel et al., 1997). Although frontal patients were comparable with healthy controls on a number of process measures, they had more difficulty structuring the multi-part
problem, and made poor judgements about how satisfactory
their proposed solutions were. Subjects with frontal damage
tended to stop before they had developed an objectively adequate plan. Similar difficulties have been noted following
frontal lobe damage in another ill-structured, real-world
task, the Multiple Errand test of Shallice and Burgess
(1991). Such studies have typically not examined D/LF and
VMF damage separately, although case reports suggest that
(relatively) isolated right D/LF/frontopolar (Goel and
Grafman, 2000), or right orbitofrontal cortex damage
(Satish et al., 1999), can be sufficient to cause difficulties in
ill-structured problem-solving tasks.
It may be that the performance of VMF subjects in the
present study is simply an expression of similar difficulties
in the context of a decision-making task. However, there are
several reasons to think that there is a more specific basis for
these findings. VMF subjects pursued a consistent strategy
(albeit one that differed from the other groups), spent as
much time and gathered as much information as other participants. Further, the demands of the multi-attribute
decision tasks administered here differed in important ways
from those of the ill-defined problem-solving tasks reviewed
above: the decision tasks were brief, the possible behaviours
were simple and relatively constrained and information was
supplied in a form intended to minimize working memory
requirements. These factors may explain why subjects with
D/LF damage performed normally in the present experiment, in contrast to the reported effects of such damage
on ill-structured problem solving. They also make it less likely
that the performance of the VMF group can be explained by
poor planning or disorganization, at least in the most general
senses of these terms.
Multi-attribute decision tasks share with ill-structured
problem solving the requirement to flexibly develop a strategic, orderly sequence of behaviours in a setting in which
the end-state is not made explicit. Participants were instructed
to make a choice, but had to specify for themselves their
950
Brain (2006), 129, 944–952
decision-making goal, and judge when it had been met. There
is growing evidence that VMF damage impairs the ability to
evaluate options. One possibility is that an underlying deficit
in the ability to compare the relative value of options led the
subjects studied here to adopt qualitatively different decisionmaking goals.
Decision making in relatively open-ended contexts can be
viewed in terms of a fundamental dichotomy between two
possible decision-making goals: seeking the best alternative
(maximizing) or seeking an acceptable alternative (satisficing). These two goals typically involve different information
search strategies. Empirical data suggest that maximizing can
be carried out with a variety of search strategies, either alternative- or attribute-based, whereas satisficing typically
involves alternative-based information acquisition, the
pattern followed by the VMF group in the present study.
Interestingly, studies of normal individuals suggest that
maximizing decision strategies are often adopted to avoid
the emotional state of regret. That is, individuals are motivated to continue to search the decision space after acceptable, or even excellent, options are identified, to avoid the
possibility of missing even better choices (Schwartz et al.,
2002; Schwartz, 2004).
It is tempting to speculate that the alternative-based information acquisition of VMF subjects in the present study
may reflect a tendency to satisfice, to judge an alternative
in absolute terms as ‘good enough’ rather than ‘seeking the
best’. This possibility requires further study, but two existing
lines of evidence are at least consistent with this hypothesis.
First, recent work using a simple gambling paradigm found
that subjects with VMF damage felt less regret after making
what turned out to be a sub-optimal choice (Camille et al.,
2004). A diminished capacity to experience regret might make
such patients less inclined to adopt a maximizing strategy.
Second, a variety of lines of evidence suggest that regions
within VMF are involved in flexibly representing the contextspecific value of stimuli more generally. Lesions of OFC in
several species, including humans, have been shown to impair
performance on tasks that require the flexible updating of
stimulus-reinforcement associations, such as reversal learning (Jones and Mishkin, 1972; Dias et al., 1996; Schoenbaum
et al., 2002; Fellows and Farah, 2003; Clark et al., 2004;
Hornak et al., 2004; Izquierdo et al., 2004). Single unit recordings in monkeys have identified OFC neurons that respond
selectively to the most highly valued stimulus offered to the
animal, and that rapidly cease responding when circumstances change in a way that reduces the value of that stimulus
(Rolls, 1999, 2000; Tremblay and Schultz, 1999; Wallis and
Miller, 2003). The somatic marker hypothesis, a prominent
theory of the role of VMF in risky decision making (Bechara
et al., 1997), is also fundamentally consistent with the view
that this region is important in tracking the value of options,
at least in certain circumstances.
There is some evidence that VMF is involved even in simple
evaluative judgements under conditions of certainty: VMF
lesions in macaques lead to abnormal food preferences
L. K. Fellows
(Baylis and Gaffan, 1991), and in humans to inconsistent
choices in a simple pairwise preference task (Fellows and
Farah, 2004). In addition, functional imaging studies of
preference judgements in healthy human subjects have implicated regions of OFC and medial prefrontal cortex in evaluation (Zysset et al., 2002; Arana et al., 2003; Cunningham et al.,
2003; Paulus and Frank, 2003; McClure et al., 2004b).
If VMF damage leads to difficulty in determining the relative value of simple stimuli, comparing the relative value of
multi-attribute choices is presumably even more difficult.
Maximizing requires determining the relative value of all
options under consideration, whereas satisficing involves judging only whether an option’s value meets some minimum
standard. As such, the tendency of VMF subjects to acquire
information in an alternative-based pattern in the present
experiment may reflect a more basic impairment in making
determinations of relative value, with a compensatory reliance
on a satisficing strategy.
The tasks used here do not provide information about
decision-making quality, so no conclusions can be drawn
about whether VMF subjects are effective in their alternative-based decision making. Future work examining this
question could shed light on whether VMF is implicated in
judgements of both absolute (good–bad) and relative
(better–worse) value, or is important only in the latter.
In summary, this process-oriented study found that VMF
damage systematically affected information acquisition in
multi-attribute decision-making problems. These findings
argue that such damage can fundamentally affect how decisions are made, even in the absence of uncertainty, risk or the
need to consider future outcomes. Process-oriented
approaches to the study of decision making would seem to
hold promise for understanding the intersection between strategic, ‘pre-decisional’ information acquisition and decision
making per se. This study suggests that the VMFs play an
important role at that intersection.
Acknowledgements
The author wishes to thank Hilary Gerstein and Alisa Padon
for technical assistance, Marianna Stark for her help in patient
recruitment, and Martha Farah and Barry Schwartz for
helpful discussion. This work was supported by NIH R21
NS045074, CIHR MOP-77583, and by a CIHR ClinicianScientist award.
References
Ackerly S. Prefrontal lobes and social development. 1950. Yale J Biol Med
2000; 73: 211–9.
Arana FS, Parkinson JA, Hinton E, Holland AJ, Owen AM, Roberts AC.
Dissociable contributions of the human amygdala and orbitofrontal
cortex to incentive motivation and goal selection. J Neurosci 2003; 23:
9632–8.
Baylis LL, Gaffan D. Amygdalectomy and ventromedial prefrontal
ablation produce similar deficits in food choice and in simple object
discrimination learning for an unseen reward. Exp Brain Res 1991; 86:
617–22.
Multi-attribute decision making following frontal damage
Bechara A, Damasio AR, Damasio H, Anderson SW. Insensitivity to future
consequences following damage to human prefrontal cortex. Cognition
1994; 50: 7–15.
Bechara A, Damasio H, Tranel D, Damasio AR. Deciding advantageously before knowing the advantageous strategy. Science 1997; 275:
1293–5.
Bechara A, Damasio H, Tranel D, Damasio AR. The Iowa Gambling Task and
the somatic marker hypothesis: some questions and answers. Trends Cogn
Sci 2005; 9: 159–62; discussion 162–4.
Berlin HA, Rolls ET, Kischka U. Impulsivity, time perception, emotion and
reinforcement sensitivity in patients with orbitofrontal cortex lesions.
Brain 2004; 127: 1108–26.
Camille N, Coricelli G, Sallet J, Pradat-Diehl P, Duhamel JR, Sirigu A. The
involvement of the orbitofrontal cortex in the experience of regret. Science
2004; 304: 1167–70.
Chu PC, Spires EE, Sueyoshi T. Cross-cultural differences in choice behavior
and use of decision aids: a comparison of Japan and the United States.
Organ Behav Hum Decis Process 1999; 77: 147–70.
Clark L, Cools R, Robbins TW. The neuropsychology of ventral
prefrontal cortex: decision-making and reversal learning. Brain Cogn 2004;
55: 41–53.
Critchley HD, Mathias CJ, Dolan RJ. Neural activity in the human brain
relating to uncertainty and arousal during anticipation. Neuron 2001;
29: 537–45.
Cunningham WA, Johnson MK, Gatenby JC, Gore JC, Banaji MR.
Neural components of social evaluation. J Pers Soc Psychol 2003; 85:
639–49.
Dias R, Robbins TW, Roberts AC. Dissociation in prefrontal cortex of
affective and attentional shifts. Nature 1996; 380: 69–72.
Ernst M, Bolla K, Mouratidis M, Contoreggi C, Matochik JA, Kurian V, et al.
Decision-making in a risk-taking task: a PET study. Neuropsychopharmacology 2002; 26: 682–91.
Eslinger PJ, Damasio AR. Severe disturbance of higher cognition after
bilateral frontal lobe ablation: patient EVR. Neurology 1985; 35:
1731–41.
Evenden JL. Varieties of impulsivity. Psychopharmacology (Berl) 1999; 146:
348–61.
Fellows LK. The cognitive neuroscience of decision making: a review and
conceptual framework. Behav Cog Neurosci Rev 2004; 3: 159–72.
Fellows LK, Farah MJ. Ventromedial frontal cortex mediates affective
shifting in humans: evidence from a reversal learning paradigm. Brain
2003; 126: 1830–7.
Fellows LK, Farah MJ. Damage to ventromedial prefrontal cortex impairs
simple preference judgments in humans. San Diego, CA: Society for
Neuroscience;, 2004.
Fellows LK, Farah MJ. Different underlying impairments in decision-making
following ventromedial and dorsolateral frontal lobe damage in humans.
Cereb Cortex 2005; 15: 58–63.
Folstein MF, Robins LN, Helzer JE. The mini-mental state examination.
Arch Gen Psychiatry 1983; 40: 812.
Ford JK, Schmitt N, Schechtman SL, Hults BM, Doherty ML. Process tracing
methods: contributions, problems, and neglected research questions. Org
Behav Hum Decision Proc 1989; 43: 75–117.
Gehring WJ, Willoughby AR. The medial frontal cortex and the rapid
processing of monetary gains and losses. Science 2002; 295: 2279–82.
Gigerenzer G, Todd PM. Simple heuristics that make us smart. New York:
Oxford University Press, 1999.
Goel V, Grafman J. Role of the right prefrontal cortex in ill-structured
planning. Cogn Neuropsychol 2000; 17: 415–36.
Goel V, Grafman J, Tajik J, Gana S, Danto D. A study of the performance of
patients with frontal lobe lesions in a financial planning task. Brain 1997;
120: 1805–22.
Harlow JM. Passage of an iron rod through the head. J Neuropsychiatry
Clin Neurosci 1999; 11: 281–3.
Hornak J, O’Doherty J, Bramham J, Rolls ET, Morris RG, Bullock PR, et al.
Reward-related reversal learning after surgical excisions in orbito-frontal
Brain (2006), 129, 944–952
951
or dorsolateral prefrontal cortex in humans. J Cogn Neurosci 2004; 16:
463–78.
Izquierdo A, Suda RK, Murray EA. Bilateral orbital prefrontal cortex lesions
in rhesus monkeys disrupt choices guided by both reward value and
reward contingency. J Neurosci 2004; 24: 7540–8.
Jones B, Mishkin M. Limbic lesions and the problem of stimulus—
reinforcement associations. Exp Neurol 1972; 36: 362–77.
Kerstholt JH. Information search and choice accuracy as a function of task
complexity and task structure. Acta Psychol 1992; 80: 185–97.
Krawczyk DC. Contributions of the prefrontal cortex to the neural
basis of human decision making. Neurosci Biobehav Rev 2002; 26:
631–64.
Maia TV, McClelland JL. A reexamination of the evidence for the somatic
marker hypothesis: what participants really know in the Iowa gambling
task. Proc Natl Acad Sci USA 2004; 101: 16075–80.
Maia TV, McClelland JL. The somatic marker hypothesis: still many
questions but no answers. Trends Cogn Sci 2005; 9: 162–4.
Manes F, Sahakian B, Clark L, Rogers R, Antoun N, Aitken M, et al. Decisionmaking processes following damage to the prefrontal cortex. Brain 2002;
125: 624–39.
McClure SM, Laibson DI, Loewenstein G, Cohen JD. Separate neural
systems value immediate and delayed monetary rewards. Science 2004a;
306: 503–7.
McClure SM, Li J, Tomlin D, Cypert KS, Montague LM, Montague PR.
Neural correlates of behavioral preference for culturally familiar drinks.
Neuron 2004b; 44: 379–87.
Newell A, Simon HA. Human problem solving. Englewood Cliffs, NJ:
Prentice-Hall, 1972.
Paulus MP, Frank LR. Ventromedial prefrontal cortex activation is critical
for preference judgments. Neuroreport 2003; 14: 1311–5.
Payne JW. Task complexity and contingent processing in decision making: An
information search and protocol analysis. Organ Behav Hum Perform
1976; 16: 366–87.
Payne JW, Bettman JR, Johnson EJ. The adaptive decision maker. New York:
Cambridge University Press, 1993.
Payne JW, Bettman JR, Coupey E, Johnson EJ. A constructive process view
of decision making: multiple strategies in judgment and choice. Acta
Psychol 1992; 80: 107–41.
Rogers RD, Everitt BJ, Baldacchino A, Blackshaw AJ, Swainson R, Wynne K,
et al. Dissociable deficits in the decision-making cognition of chronic
amphetamine abusers, opiate abusers, patients with focal damage to
prefrontal cortex, and tryptophan-depleted normal volunteers: evidence
for monoaminergic mechanisms. Neuropsychopharmacology 1999; 20:
322–39.
Rolls ET. The brain and emotion. Oxford: Oxford University Press, 1999.
Rolls ET. The orbitofrontal cortex and reward. Cereb Cortex 2000; 10:
284–94.
Rorden C, Brett M. Stereotaxic display of brain lesions. Behav Neurol 2000;
12: 191–200.
Sanfey AG, Hastie R, Colvin MK, Grafman J. Phineas gauged: decisionmaking and the human prefrontal cortex. Neuropsychologia 2003; 41:
1218–29.
Satish U, Streufert S, Eslinger PJ. Complex decision making after
orbitofrontal damage: neuropsychological and strategic management
simulation assessment. Neurocase 1999; 5: 355–64.
Schoenbaum G, Nugent SL, Saddoris MP, Setlow B. Orbitofrontal lesions
in rats impair reversal but not acquisition of go, no-go odor discriminations. Neuroreport 2002; 13: 885–90.
Schwartz B. The tyranny of choice. Sci Am 2004; 290: 70–5.
Schwartz B, Ward A, Monterosso J, Lyubomirsky S, White K, Lehman DR.
Maximizing versus satisficing: happiness is a matter of choice. J Pers Soc
Psychol 2002; 83: 1178–97.
Shallice T, Burgess PW. Deficits in strategy application following frontal
lobe damage in man. Brain 1991; 114: 727–41.
Simon D, Krawczyk DC, Holyoak KJ. Construction of preferences by
constraint satisfaction. Psychol Sci 2004; 15: 331–6.
952
Brain (2006), 129, 944–952
Simon HA. A behavioral model of rational choice. Q J Econ 1955; 69:
99–118.
Stuss DT, Levine B. Adult clinical neuropsychology: lessons
from studies of the frontal lobes. Annu Rev Psychol 2002; 53:
401–33.
Sugrue LP, Corrado GS, Newsome WT. Choosing the greater of two goods:
neural currencies for valuation and decision making. Nat Rev Neurosci
2005; 6: 363–75.
Tanaka SC, Doya K, Okada G, Ueda K, Okamoto Y, Yamawaki S. Prediction
of immediate and future rewards differentially recruits cortico-basal
ganglia loops. Nat Neurosci 2004; 7: 887–93.
L. K. Fellows
Tomb I, Hauser M, Deldin P, Caramazza A. Do somatic markers mediate
decisions on the gambling task? Nat Neurosci 2002; 5: 1103–4; author reply
1104.
Tremblay L, Schultz W. Relative reward preference in primate orbitofrontal
cortex. Nature 1999; 398: 704–8.
Wallis JD, Miller EK. Neuronal activity in primate dorsolateral and orbital
prefrontal cortex during performance of a reward preference task. Eur J
Neurosci 2003; 18: 2069–81.
Zysset S, Huber O, Ferstl E, von Cramon DY. The anterior frontomedian
cortex and evaluative judgment: an fMRI study. Neuroimage 2002; 15:
983–91.