Cerebral Cortex July 2007;17:1625--1636
doi:10.1093/cercor/bhl073
Advance Access publication September 8, 2006
Thalamic--Prefrontal Cortical--Ventral
Striatal Circuitry Mediates Dissociable
Components of Strategy Set Shifting
Annie E. Block, Hasina Dhanji, Sarah F. Thompson-Tardif and
Stan B. Floresco
The mediodorsal nuclei of thalamus (MD), prefrontal cortex (PFC),
and nucleus accumbens core (NAc) form an interconnected network that may work together to subserve certain forms of
behavioral flexibility. The present study investigated the functional
interactions between these regions during performance of a crossmaze--based strategy set-shifting task. In Experiment 1, reversible
bilateral inactivation of the MD via infusions of bupivacaine did
not impair simple discrimination learning, but did disrupt shifting
from response to visual cue discrimination strategy, and vice versa.
This impairment was due to an increase in perseverative errors.
In Experiment 2, asymmetrical disconnection inactivations of
the MD on one side of the brain and PFC on the other also caused
a perseverative deficit when rats were required to shift from a
response to a visual cue discrimination strategy, as did disconnections between the PFC and the NAc. However, inactivation
of the MD on one side of the brain and the NAc contralaterally
resulted in a selective increase in never-reinforced errors, suggesting this pathway is important for eliminating inappropriate
strategies during set shifting. These data indicate that set shifting
is mediated by a distributed neural circuit, with separate neural
pathways contributing dissociable components to this type of
behavioral flexibility.
primates plays an essential role in reversal learning, a simpler
form of behavioral flexibility, whereas lesions to the medial PFC
in rodents or the dorsolateral PFC in primates are without effect
(Dias and others 1997; Birrell and Brown 2000; McAlonan and
Brown 2003; Kim and Ragozzino 2005).
It is becoming increasingly apparent that different subcortical
afferents of the PFC contribute to behavioral flexibility mediated by the frontal lobes. For example, both the dorsal and
ventral regions of the striatum have been implicated in different
forms of set shifting. Patients with pathology of the dorsal
striatum display impairments in extradimensional shifts (Owen
and others 1993; Lawrence and others 1999), and functional
imaging studies have shown increased activation of the caudate
nucleus specifically during the reception of negative feedback
(Monchi and others 2001). In keeping with these findings,
inactivation of the dorsomedial striatum in rats disrupts shifting
from one discrimination strategy to another, impairing selectively the ability to maintain new strategies (Ragozzino and
others 2002). Recent findings in our laboratory have also
implicated the nucleus accumbens core (NAc) in strategy set
shifting. The NAc receives a dense projection from the medial
PFC (Brog and others 1993; Gabbott and others 2005), and
inactivation of this nucleus also impairs strategy set shifting in
a manner distinct from that observed following manipulations of
the PFC or dorsal striatum. Inactivation of the NAc (but not
shell) caused a pattern of errors that indicated a failure to
maintain new strategies as well as elimination of inappropriate
response alternatives during a strategy shift (Floresco, GhodsSharifi, and others 2006), similar to ‘‘loss of set’’ errors, made by
Parkinsonian patients performing a set shift (Gauntlett-Gilbert
and others 1999). Thus, impairments induced by inactivation of
the dorsomedial striatum or NAc are qualitatively different from
the perseverative impairment one sees with manipulations of
the PFC, indicating that these regions may play functionally
distinct roles in the shifting of behavioral strategies.
Another subcortical region that shares projections with
both the PFC and the NAc are the mediodorsal nuclei of the
thalamus (MD). The PFC is reciprocally connected with MD
(Groenewegen 1988; Condé and others 1995), and afferents
from the MD also terminate in the NAc (Berendse and
Groenewegen 1990). Given this anatomical connectivity, it is
not surprising that this diencephalic region has also been
implicated in different types of behavioral flexibility. Functional
imaging studies have revealed that performance of the WCST
causes significant activation of the MD during the receipt of
negative feedback, indicating that this region of the thalamus
may flag the need to initiate an adaptive strategy shift (Monchi
and others 2001). Schizophrenic individuals also show deficits
in strategy set shifting (Pantelis and others 1999), which has
Keywords: executive function, frontal lobe, mediodorsal thalamus,
nucleus accumbens
Introduction
It is well established that different subregions of the mammalian
prefrontal cortex (PFC) mediate the ability to adapt behavior in
response to changing environmental contingencies. For example, attentional set shifting is a type of behavioral flexibility
that requires an organism to disengage from a once relevant set
of stimulus dimensions and begin responding to a previously
irrelevant set to perform optimally. In humans, this executive
function is assessed using the Wisconsin card sort task (WCST),
a neuropsychological task that is critically dependent on the
integrity of the dorsolateral PFC in humans (Lombardi and
others 1999). Patients with damage to the PFC acquire an initial
sorting rule readily during performance of the WCST, but when
the rule is shifted without warning, patients fail to adapt their
responding to the negative feedback (Stuss and others 2000).
Similarly, lesions of the dorsolateral PFC in primates or the
medial PFC in rats impair set shifting on an intradimensional/
extradimensional shifting task (Dias and others 1997; Birrell
and Brown 2000). Moreover, manipulation of the medial PFC
in rats also impairs the ability to shift from one discrimination
strategy to another using a maze-based set-shifting procedure
(Ragozzino and others 1999; Stefani and others 2003; Floresco,
Magyar, and others 2006). In contrast, the orbital PFC in rats and
The Author 2006. Published by Oxford University Press. All rights reserved.
For permissions, please e-mail: [email protected]
Department of Psychology and Brain Research Center,
University of British Columbia, Vancouver, British Columbia,
Canada V6T 1Z4
been linked to the observed loss of cells from the MD in
schizophrenic individuals relative to controls (Popken and
others 2000). Additionally, Korsakoff’s amnesiacs, whose lesions
encompass the MD (Joyce 1987), show perseverative impairments on the WCST (Oscar-Berman and others 2005). Studies in
experimental animals have also implicated the MD as an important nucleus for regulating different forms of behavioral
flexibility. For instance, lesions of the MD disrupt reversal
learning under certain conditions, although these effects are
variable and are dependent on a number of factors. These
include the type of stimuli used or whether the animal is subjected to a single or multiple reversals (Tigner 1974; Means and
others 1975; Chudasama and others 2001; but see Beracochea
and others 1989). However, neurotoxic lesions of the MD cause
a robust increase in perseverative-type errors when rats were
required to shift response rules from a match to a nonmatch
to place strategy on a working memory task, in a manner similar
to rats with lesions of the medial PFC (Hunt and Aggleton 1998;
Dias and Aggleton 2000). This latter finding indicates that the
MD may be particularly important in situations that require
strategy shifting.
The data reviewed above suggest that the MD is part of a
neural circuit that facilitates behavioral flexibility mediated in
part by the PFC and NAc. However, the specific contribution
that the MD makes to this executive function remains to be
explored fully. Furthermore, given that the MD projects to both
the PFC and the NAc, it is unclear whether this nucleus interacts
with one or both of these regions to mediate behaviors such
as strategy set shifting. Thus, in Experiment 1, we investigated
the role of the MD in strategy set shifting known to be mediated
by the medial PFC and NAc (Ragozzino and others 1999;
Floresco, Ghods-Sharifi, and others 2006). Rats with reversible
bilateral inactivations of the MD performed 1 of 2 strategy set
shifts: either the acquisition of an egocentric response and
a subsequent shift to a visual cue--based strategy or the reverse.
In Experiment 2, we used reversible asymmetrical disconnection lesions to investigate the possible routes of serial information transfer between the MD, NAc, and PFC during set
shifting.
Experiment 1A and B
Materials and Methods
Subjects
Male Long-Evans rats (Charles River Laboratories, Montreal, Canada)
weighing 280--360 g at the beginning of the experiment were used. Rats
were individually housed in plastic cages in a temperature controlled
room (20 C) on a 12-h light--dark cycle. Immediately following surgery,
all rats were restricted to 85% of their free-feeding weight, with free
access to water for the duration of the experiment.
Apparatus
A four-arm cross-maze was used, made of 1.5 cm thick plywood and
painted white. Each arm was 60 cm long and 10 cm wide, with 20 cm
high walls on each arm and with cylindrical food wells (2 cm wide 3 1
cm deep) drilled into the end of each of the arms, 2 cm from the end
wall. Four table legs attached to the ends of each arm elevated the maze
70 cm above the floor. Removable pieces of white opaque plastic (20 3
10 cm) were used to block the arms of the maze to form a ‘‘T’’
configuration. The maze resided in a room measuring 3.4 3 3.4 m.
Surgery
Rats were anesthetized with 100 mg/kg xylazine and 7 mg/kg ketamine
hydrochloride and implanted with 23 gauge bilateral stainless steel
1626 Thalamic--Prefrontal Circuits and Set Shifting
d
Block and others
guide cannulae into the MD (flat skull—anterior--posterior [AP]: –2.9 mm
from bregma, medial--lateral [ML]: ±0.7 mm from midline, dorsal--ventral
(DV): –4.9 mm from dura). The coordinates were based on the atlas of
Paxinos and Watson (1998). Four jeweler’s screws were implanted
surrounding the cannulae and secured in place with dental acrylic.
Thirty gauge obdurators flush with the end of the guide cannulae
remained in place until the rats were given infusions. Each rat was given
at least 7 days to recover from surgery prior to behavioral training.
During this recovery period, animals were food restricted and were
handled for at least 5 min per day.
Microinfusion Procedure
Infusions into the MD were made through 30 gauge injection cannulae
extending 0.8 mm below the guide cannulae. The injection cannulae
were attached by a polyethylene tube to a 10-lL syringe. Saline or the
local anesthetic bupivacaine hydrochloride (0.75%; Abbott Laboratories
Saint Laurent, Quebec, Canada) was infused at a rate of 0.5 ll per 72 s by
a microsyringe pump (Sage Instruments Model 341). Injection cannulae
were left in place for an additional 1 min to allow for diffusion. Each rat
remained in its home cage for a further 10 min prior to behavioral
testing.
Ten minutes before both the response discrimination learning and the
strategy set shift, rats received a microinfusion. For both Experiments 1A
and B, each rat was assigned to 1 of 3 infusion treatment groups, 1) Day
1-saline and Day 2-saline, 2) Day 1-saline and Day 2-bupivacaine, and 3)
Day 1-bupivacaine and Day 2-saline. Group 1 served as the control
group, group 2 determined whether MD inactivation impaired strategy
set shifting, and group 3 determined whether inactivation of the MD
would affect the acquisition or consolidation of discrimination learning
that may yield behavioral differences during the set shift.
Maze Familiarization Procedure
The familiarization and strategy set-shifting procedures have been
described previously (Ragozzino 2002; Floresco, Ghods-Sharifi, and
others 2006; Floresco, Magyar, and others 2006). Before the first day
of familiarization to the maze, the rats were pre-exposed to 10--20 of the
reward pellets they would receive in the maze. On the first day of
familiarization, rats were placed in the center of the cross-maze, which
had each arm baited with 5 pellets: 2 in each well and 3 down the length
of the arm. A rat was placed in the maze and allowed to freely navigate
and consume the food pellets for 15 min. If a rat consumed all 20 pellets
prior to 15 min, it was removed from the maze and placed in a holding
cage, the maze was rebaited with 20 additional pellets, and the rat was
placed back in the center of the maze. The second day of familiarization
was the same as the first day, except there were only 3 pellets in each
arm: 2 in each well, and 1 on each arm. On subsequent days of maze
familiarization, only 4 pellets were placed on the maze, 1 in each well.
Additionally, a black laminated piece of poster board (9 3 20 3 0.3 cm),
which served as the visual cue, was placed in a random arm and rotated
between rebaitings of the maze. Rats were picked up and placed at the
start of an arm, allowed to traverse the arm of the maze, consume the
pellet, and were immediately picked up and placed at the beginning of
another baited arm. Rats continued this procedure daily until they
depleted the maze 4 or more times in 15 min. Rats required an average
of 4.58 ±0.3 days of familiarization (range, 3--14 days) to reach this
criterion.
After the rat had achieved familiarization criterion, the turn bias for
the rat was determined. The white opaque plexiglas insert was placed at
the entrance of one of the arms, forming a ‘‘T’’ configuration. A rat was
placed in the stem arm and allowed to turn left or right to obtain a food
pellet. In one of the choice arms, the visual cue insert was placed on the
floor. After a rat chose an arm and consumed a food pellet, it was picked
up, placed in the stem arm, and allowed to make the next choice. If the
rat chose the same arm as the initial choice, it was returned to the stem
arm until it chose the remaining arm and consumed the food pellet.
After choosing both arms, the rat was returned to the holding cage, the
Plexiglas barrier and visual cue were moved to different arms, and a new
trial commenced. Thus, a trial for the turn-bias procedure consisted of
entering both choice arms and consuming the food pellets. The turn
that a rat made first during the initial choice of a trial was recorded and
counted toward its turn bias, and the direction (right or left) that a rat
turned 4 or more times over 7 total trials was considered its turn bias.
After determining the turn bias, a rat’s obdurators were removed from
the guide cannulae and 2 injection cannulae extending 0.8 mm past the
guide cannulae were inserted for 1 min, but no solution was injected at
this time. This procedure was performed to familiarize the animal to the
2 infusions they would receive over the next 2 days of testing. Response
(Experiments 1A and 2) or visual cue (Experiment 1B) discrimination
training commenced on the following day.
Experiment 1A, Day 1: Response Discrimination Learning
A detailed discussion of the merits of this type of set-shifting task
and comparisons with other set-shifting tasks used in rodents has
been addressed previously (Floresco, Ghods-Sharifi, and others 2006;
Floresco, Magyar, and others 2006). For this discrimination, the rat was
required to always turn in the direction opposite its turn bias (left or
right), regardless of the location of the visual cue placed in one of the
arms (See Fig. 1A, upper panel). Over the course of training, 1 of 3 start
arms were used, discouraging the use of an allocentric spatial strategy to
locate the food. On Day 1 of training, a rat was started from the arms
designated west, south, and east (W, S, and E, respectively). The location
of these arms relative to the spatial cues in the room was varied across
Figure 1. Example of the strategy set-shifting task used in Experiments 1A and 2. The
arrows in the maze represent the correct navigation pattern to receive reinforcement.
(A) During initial response discrimination training on Day 1 (upper panels), in this
example, the rat was started from the south (S), west (W), and east (E) arms and
always had to make a 90 turn to the right to receive food reinforcement. A black and
white striped visual cue was randomly placed in one of the choice arms on each trial
but did not reliably predict the location of food during response training. During the set
shift on Day 2 (lower panels), the rat is required to use a visual cue discrimination
strategy. Here, the rat was started from the same arms but had to always enter the
arm with visual cue, which could require either a right or left turn. Thus, the rat must
shift from the old strategy and approach the previously irrelevant cue in order to obtain
reinforcement. (B) Examples of the 3 types of errors that rats could make during the
set shift. See Materials and Methods for details.
animals, so that the maze was placed in 1 of 4 possible orientations. For
every trial, the visual cue was placed in one of the choice arms so that
over every consecutive set of 12 trials it was placed an equal number of
times in each choice arm, with no more than 2 appearances in a row in
any one arm. The order of the start location for each trial, as well as the
position of the visual cue, was determined pseudorandomly and taken
from a preset sequence that was identical for each animal. On an
individual trial, the rat was placed in the stem arm and required to make
the appropriate turn in order to receive a food pellet. Between trials,
a rat was placed back in the holding cage on a bench adjacent to the
maze. The intertrial interval was ~15 s. A rat continued to receive
training trials until it reached a criterion of 10 correct consecutive
choices. There was no limit on the number of trials a rat was allotted to
reach this criterion. After the rat achieved this acquisition criterion, it
received a probe trial; this consisted of starting the rat from the fourth
arm (north, N) that was never a start arm prior to acquisition criterion.
During probe trials, the visual cue was inserted in the arm opposite to
the direction that the rat was required to turn. If a rat correctly turned
the same direction as was required during training, then response
discrimination training was completed. If a rat made an incorrect turn,
response training was continued until a rat made an additional 5 correct
choices consecutively, at which time another probe trial was administered. This procedure was continued until a rat made a correct choice
on the probe trial. The following measures were taken for each rat and
used for data analysis: 1) trials to criterion, defined as the total number of
test trials completed before a correct choice on the probe trial was
made and 2) probe trials, defined as the total number of probe trials an
animal required to get one correct. The total time it took to complete
training was also recorded.
Experiment 1A, Day 2: Shift to Visual Cue
The acquisition of this discrimination required the animal to cease the
use of a response strategy and instead use a visual cue--based strategy to
obtain food reward (Fig. 1A, lower panel). The day following successful
acquisition of the response discrimination, rats were trained to enter
the arm that contained the visual cue, the location of which was
pseudorandomly varied in the left and right arms such that it occurred in
each arm with equal frequency for every consecutive set of 12 trials. The
same training procedure, start arms and criteria to complete the visual
cue version, were used as described in the response version. For probe
trials, the visual cue was placed in the arm opposite that the rat had been
trained to enter during response discrimination training.
Errors committed during the set shift were broken down into 3 error
subtypes to determine whether treatments altered the ability to either
shift away from the previously learned strategy (perseverative errors) or
acquire and maintain the new strategy after perseveration had ceased
(regressive or never-reinforced errors). A ‘‘perseverative error’’ was
scored when a rat made an egocentric response as required on the
response version (e.g., turned right) on trials that required the rat to
turn in the opposite direction to enter the arm containing the visual cue
(Fig. 1B, left panel). Six of every 12 consecutive trials required the rat to
respond in this manner (i.e., enter the arm opposite of the previously
learned turn direction). As described in previous studies (Dias and
Aggleton 2000; Ragozzino and others 1999; Ragozzino 2002; Floresco,
Ghods-Sharifi, and others 2006; Floresco, Magyar, and others 2006),
these types of trials were separated into consecutive blocks of 4 trials
each. Perseverative errors were scored when a rat entered the incorrect
arm on 3 or more trials per block of 4 trials where the rat was required
to enter the arm opposite of the previously learned turn direction. Once
a rat made less than 3 perseverative errors in a block for the first time, all
subsequent errors of this type were no longer counted as perseverative
errors because at this point the rat was choosing an alternative strategy
at least half of the time. Instead, these errors were now scored as
‘‘regressive’’ errors (Fig. 1B, left panel). The third type of error, termed
‘‘never-reinforced errors,’’ was scored when a rat entered the incorrect
arm on trials where the visual cue was placed in the same arm that the
rat had been trained to enter on the previous day (Fig. 1B, right panel).
For example, during training on Day 1, a rat might be required to turn
right. During the shift on Day 2, a rat must now enter the arm with the
visual cue, and for half of the trials, the cue was in the right arm. In this
situation, a never-reinforced error was scored when a rat entered the
Cerebral Cortex July 2007, V 17 N 7 1627
left arm (i.e., a choice that was not reinforced on either Day 1 or Day 2).
Regressive and never-reinforced errors are used as an index of the
animals’ ability to maintain and acquire a new strategy, respectively.
Experiment 1B: Visual Cue to Response Set Shift
For these experiments, rats were initially trained on the visual cue
version of the task on Day 1 followed by testing on the response version
on Day 2. All other aspects of the testing procedure were identical to
those described above. On the shift to the response version, the same
measures were assessed as those for Experiment 1, where rats were
required to shift from a response to a visual cue strategy. However,
perseverative and regressive errors were analyzed from the trials in
which a rat was required to turn in the arm opposite to that of the visual
cue. Ten minutes before each test day, rats received a microinfusion of
either saline or bupivacaine, as described above.
Histology
Upon completion of behavioral testing, the rats were sacrificed in
a carbon dioxide chamber. Brains were removed and fixed in a 4%
formalin solution. The brains were frozen and sliced in 50-lm sections
prior to being mounted and stained with cresyl violet. Placements were
verified with reference to the neuroanatomical atlas of Paxinos and
Watson (1998).
Statistical Analysis
Separate analyses of variance (ANOVAs) were conducted on trials to
criterion data from the response acquisition on Day 1. Trials to criterion
data obtained from Day 2 were analyzed using a 2-way between/withinsubjects ANOVA, with treatment as the between-subjects factor and
choice Type as the within-subjects factor. Multiple comparisons were
conducted using Dunnett’s test. Trials per minute, as well as the number
of probe trials required to reach criterion, were analyzed using separate
1-way ANOVAs.
Results
Experiment 1A: Effects of Bilateral MD Inactivation on a
Response to Visual Cue Set shift
Histology. Figure 2 illustrates the location of the cannula tips in the MD.
All animals with placements encroaching on habenula or subthalamic
nuclei were excluded. Rats receiving inactivations of the habenula were
unimpaired on the strategy set shift, though those animals with
subthalamic placements generally took more trials to criterion than
saline-treated animals.
Day 1: response discrimination. The results for the acquisition of the
response strategy on Day 1 are shown in Figure 3A. There were no
significant difference between groups in total trials to reach the
criterion (F2,17 = 3.11, not significant [NS]). Additionally, neither there
were any differences in the number of probe trials required to reach the
criterion (F2,17 = 0.14, NS) nor were there differences in the number of
trials completed per minute between groups (F2,17 = 0.07, NS). Thus,
inactivation of the MD does not impair learning of a simple response
discrimination.
Day 2: shift to visual cue. The results for the set shift to visual cue
strategy are shown in Figure 3B. This analysis revealed a significant main
effect of treatment (F2,17 = 14.27, P < 0.001). Additionally, there was
a significant main effect of choice type (F1,17 = 87.98, P < 0.001), as rats
made significantly more correct choices than errors over the course of
the task. There was no choice 3 treatment interaction (F2,17 = 0.36, NS).
Dunnett’s test showed that rats receiving inactivation of the MD before
the set shift on Day 2 (n = 7) took significantly more trials to reach
criterion than rats receiving saline (n = 7, P < 0.05), whereas rats
receiving inactivations on Day 1 did not differ from controls (n = 6).
The number of errors committed during the strategy set shift was
analyzed separately using a mixed between/within-subjects ANOVA,
with treatment as the between-subjects factor and the 3 types of errors
measured as the within-subjects factors. A main effect for treatment
1628 Thalamic--Prefrontal Circuits and Set Shifting
d
Block and others
Figure 2. Schematic of coronal sections of the rat brain showing the placements of
the cannulae tips for all rats that received infusions of bupivacaine into the MD. Brain
sections adapted from Paxinos and Watson (1998). Numbers correspond to
millimeters from bregma.
(F2,17 = 13.69, P < 0.001) and error type was observed (F2,32 = 16.00, P <
0.001), as well as a significant error 3 treatment interaction (F4,34 = 4.35,
P < 0.01). Simple main effects analyses and multiple comparisons
revealed that this interaction was due to the fact that inactivation of the
MD prior to the set shift increased the number of perseverative errors
relative to controls (F2,17 = 9.56, P < 0.005 and Dunnett’s, P < 0.001; see
Fig. 3C). Rats receiving inactivations of the MD on Day 1 and saline on
Day 2 were indistinguishable from control rats. Furthermore, there was
a noticeable increase in never-reinforced errors in rats receiving
inactivations of the MD prior to the set shift, although this effect did
not achieve statistical significance (F2,17 = 1.65, NS). There was no
significant difference between groups in the number of regressive
errors (F2,17 = 0.29, NS). Additional analyses of the number of probe trials
required before reaching criterion and the number of trials completed
per minute (see Table 1) revealed no significant differences between
groups (all F values < 1.02, NS). Thus, inactivations of the MD caused an
increase in perseverative errors on this type of set-shifting task, in
a manner similar to PFC manipulations (Ragozzino and others 1999;
Stefani and others 2003; Floresco, Magyar, and others 2006).
Experiment 1B: Effects of MD Inactivation on a Visual Cue to
Response Set Shift
Day 1: visual cue discrimination. The results for the acquisition of the
visual cue--based strategy on Day 1 are shown in Figure 3D. A 1-way
ANOVA showed no significant difference between groups in total trials
to reach the criterion (F2,20 = 0.182, NS). Additionally, neither there
were any differences found in the number of probe trials required to
reach the criterion (F2,20 = 0.13, NS) nor were there differences in the
number of trials taken per minute between groups (F2,20 = 0.50, NS).
Figure 3. Experiment 1: Bilateral inactivation of the MD disrupts shifting from a response to a visual cue--based strategy (A--C) and vice versa (D--F). Data are expressed as means +
standard error of mean. (A) Trials to criterion on acquisition of a response discrimination on Day 1 by rats receiving infusions of saline (white, black bars) or bupivacaine into the
MD (hatched bar). Inactivation of the MD did not impair response learning. (B) Trials to criterion on the shift to visual cue discrimination strategy on Day 2 following infusions of
either saline (white and hatched bar) or bupivacaine into the MD (black bar; single star denotes P < 0.05 significantly different from saline/saline). (C) Analysis of the type of errors
committed in Experiment 1A during the set shift on Day 2. Inactivation of the MD prior to the set shift (black bars) significantly increased the number of perseverative errors (left) but
did not affect regressive (middle) and never-reinforced (right) errors (double stars denote P < 0.01 significantly different from saline/saline, white bars). (D) Trials to criterion on
acquisition of the visual cue discrimination on Day 1 by rats receiving either infusions of saline (white, black bars) or bupivacaine into the MD (hatched bar). Inactivation of the MD
did not impair visual cue discrimination learning. (E) Trials to criterion on the shift to the response discrimination on Day 2 following infusions of either saline (white and hatched bars)
or bupivacaine into the MD (black bar; double stars denote P < 0.01 significantly different from saline/saline). (F) Again, MD inactivation had a significant effect on perseveration
(left), and in this case, never-reinforced errors (right), but did not increase the number of regressive errors (middle) during the shift on Day 2 (double stars denote P < 0.01, single
star denotes P < 0.05 significantly different from the same type of errors made by saline/saline-treated rats).
Table 1
Means ± standard error of mean trials completed per minute for Experiments 1 and 2
Day 1
Day 2
Experiment 1A (response to cue)
Experiment 1B (cue to response)
Experiment 2 (response to cue)
Saline/saline
Bupi/saline
Saline/bupi
Saline/saline
Bupi/saline
Saline/bupi
Saline
Unilateral
Ipsilateral
PFC--NAc
MD--NAc
MD--PFC
1.33 ± 0.07
1.37 ± 0.10
1.29 ± 0.10
1.56 ± 0.07
1.24 ± 0.15
1.46 ± 0.14
1.72 ± 0.14
1.66 ± 0.08
1.60 ± 0.07
1.50 ± 0.07
1.75 ± 0.12
1.72 ± 0.10
1.60 ± 0.09
1.83 ± 0.11
1.52 ± 0.11
1.69 ± 0.10
2.23 ± 0.19
2.43 ± 0.25
1.48 ± 0.22
2.00 ± 0.09
1.59 ± 0.10
1.51 ± 0.07
1.89 ± 0.11
2.10 ± 0.10
Note: bupi, bupivacaine.
Thus, inactivation of the MD does not impair the acquisition of a simple
visual cue--based discrimination. One rat receiving bupivacaine on Day 1
and saline on Day 2 was excluded from the analysis as a statistical outlier,
with its trials to reach criterion on Day 2 more than 2 standard deviation
(SD) away from the mean of the group with that data point inclusive
(mean = 52.0, SD = 34.18; excluded rat = 180 trials).
Day 2: shift to response. The results from the shift to a response-based
strategy are shown in Figure 3E. The analysis of these data revealed
a significant main effect of treatment (F2,20 = 4.56, P < 0.05), a main
effect of choice (F2,20 = 116.66, P < 0.001). Dunnett’s test indicated that
rats receiving inactivations of the MD before the set shift on Day 2 (n =
8) took significantly more trials to reach criterion than rats receiving
saline (P < 0.01, n = 8), though rats receiving Day 1 inactivation and Day
2 saline did not differ from controls (n = 7).
Analysis of the errors committed during the set shift revealed
a significant main effect of treatment (F2,20 = 4.73, P < 0.05), error
type (F4,40 = 31.72, P < 0.001), and error 3 treatment interaction (F2,40 =
3.29, P < 0.05, Fig. 3F). Subsequent 1-way ANOVAs for each error type
revealed no significant treatment effect for regressive errors (F2,20 =
0.57, NS). However, as in Experiment 1, a significant treatment effect
was observed for perseverative errors (F2,20 = 4.84, P < 0.05). Pairwise
comparisons indicated that this was due to an increase in perseverative
errors for animals receiving MD inactivations prior to the set shift on
Day 2 relative to controls (P < 0.005), and animals receiving Day 1
inactivations did not differ from controls. In this experiment, we also
observed a significant main effect of treatment for never-reinforced
errors for this strategy set shift (F2,20 = 3.71, P < 0.05). Additional
analyses of the number of trials completed per minute and the number
of probe trials required to complete the strategy set shift show no
significant treatment effects (all F values < 2.42, NS). Thus, Experiment
1B both confirms the results of Experiment 1A and indicates that the MD
may play a role in the generation of alternative strategies, as evidenced
by the increase in never-reinforced errors.
Cerebral Cortex July 2007, V 17 N 7 1629
Experiment 2: Effects of Asymmetrical Disconnection
Inactivations of the PFC, NAc, and MD on a Response
to Visual Cue Set Shift
The results from Experiment 1 indicate a role for the MD in set shifting.
However, the MD sends projections to both PFC and NAc, which
have been implicated in mediating this form of behavioral flexibility
(Ragozzino and others 1999; Floresco, Ghods-Sharifi, and others 2006;
Floresco, Magyar, and others 2006). In Experiment 2, we use an asymmetrical disconnection inactivation procedure to delineate the routes of
information transfer within this thalamic--cortical--striatal circuit. Disconnection lesions are used to identify functional components of a
circuit where information is transferred serially from one structure to
another structure in the same hemisphere, on both sides of the brain in
parallel. Furthermore, the design assumes that dysfunction will result
from blockade of neural activity at the origin of a pathway in one
hemisphere and the termination of the efferent pathway in the contralateral hemisphere. It follows that a unilateral inactivation at either
site should have no effect on behavior, as the unaffected brain regions in
the contralateral hemisphere will serve to compensate. We have used
this reversible disconnection procedure previously to delineate the
routes of information transfer between brain regions for both working
memory and decision making processes (Floresco and others 1997,
1999; Floresco and Ghods-Sharifi 2006). As inactivation of the MD, NAc
(Floresco, Magyar, and others 2006), or PFC (Ragozzino and others
1999) disrupted performance on both response to visual cue and visual
cue to response set shifts while having no effect on the initial learning
of either discrimination, in this experiment, we only tested animals
performing a response to visual cue set shift.
Materials and Methods
Subjects and Surgery
Three groups of male Long-Evans rats (290--380 g) were implanted with
2 sets of bilateral cannula. One group of rats was implanted with one pair
of cannula in the MD and a second pair in the prelimbic region of the
PFC (flat skull: AP = +3.0 mm, ML = ± 0.7 mm from bregma, and DV = –2.7
mm from dura). A second group was implanted with one pair of cannula
in the MD and a second pair in the NAc (flat skull: AP = +1.6 mm, ML = ±
1.8 mm from bregma, and DV = –6.0 mm from dura). A third group of rats
was implanted with one pair of cannula in the PFC and a second pair in
the NAc. The cannulation into the NAc at these coordinates transects
the corpus callosum, effectively eliminating any contralateral projections from the PFC to the NAc (Gorelova and Yang 1997). This point is
important as disconnection lesions are most effective where there are
no contralateral projections between brain regions (Floresco and others
1999; Dunnett and others 2005). Thirty gauge obdurators flush with the
end of the guide cannulae remained in place until the injections were
made. Each rat was given at least 7 days to recover from surgery before
testing. The location of acceptable infusion placements in all brain
regions tested in Experiment 2 is presented in Figure 4.
Maze Familiarization Procedure and Response to Visual Cue
Set-Shifting Procedure
The familiarization response discrimination training and set-shifting
procedures were identical to that used in Experiment 1A.
Microinfusion Procedure
All animals received mock infusions (injection cannula inserted with no
flow) following the turn bias procedure on the last day of maze
familiarization, as well as before response discrimination training on
Day 1. The day after acquisition of the response discrimination, and
before the set shift on Day 2, all animals received injections in the
opposite hemisphere of that which received the mock infusion
the previous day. Unilateral inactivations of MD and PFC were induced
with microinfusions of bupivacaine as in Experiment 1. However, for
inactivations of the NAc, we used the c-aminobutyric acid (GABA) agonists baclofen and muscimol (Sigma-Aldrich Canada, Ontario, Canada)
(75 ng/lL of each drug dissolved in saline and infused at a volume of
0.3 lL). This procedure was employed because projections from brain
regions that terminate in the NAc shell pass through the NAc (Gorelova
and Yang 1997). The use of GABA agonists inactivates cell bodies in the
NAc but leave fibers of passage relatively intact (McFarland and Kalivas
2001; Floresco, Ghods-Sharifi, and others 2006).
In total, 6 groups of rats were tested, 3 of which received
disconnection treatments as follows: 1) a unilateral bupivacaine infusion
Figure 4. Schematic of coronal sections of the rat brain showing the placements of the cannulae tips for all rats that received double cannulations. Numbers beside each plate
correspond to millimeters from bregma. (A) Location of cannulae tips (black circles) for all rats used for data analysis receiving MD--PFC infusions, (B) PFC--NAc infusions or (C) MD-NAc infusions. For clarity, figures represent the asymmetrical infusion procedure; all rats received infusions in either the left or the right hemisphere that was counterbalanced
across animals.
1630 Thalamic--Prefrontal Circuits and Set Shifting
d
Block and others
into the PFC in combination with bupivacaine infused into the
contralateral MD, 2) a unilateral bupivacaine infusion into the PFC in
combination with a contralateral infusion of baclofen/muscimol into the
NAc, and 3) a unilateral bupivacaine infusion into the MD in combination with a contralateral infusion of baclofen/muscimol into the NAc.
Two initial control groups were formed, one group received unilateral
inactivations with a saline infusion into the contralateral structure as
follows: 1) a unilateral baclofen/muscimol into NAc/saline into the
contralateral MD or PFC, 2) unilateral bupivacaine into MD/saline into
NAc or PFC, and 3) unilateral bupivacaine into the PFC/saline into the
contralateral NAc or MD. The saline control group was composed of the
following combinations of asymmetrical bilateral infusions: 1) unilateral
infusions of saline into both MD and NAc, 2) unilateral infusions of saline
into both PFC and NAc, and 3) unilateral infusions of saline into the MD
and PFC (See Table 2). The hemisphere (left or right) used for the
injection was also counterbalanced across animals within groups. With
respect to combined ipsilateral inactivation control treatments, numerous studies using disconnection designs have shown that ipsilateral
lesions of 2 interconnected structures in the same hemisphere do not
impair behavior relative to the effect of crossed, disconnection lesions
(Olton and others 1982; Warburton and others 2000; Chudasama and
others 2003; Dunnett and others 2005). Nevertheless, because PFC--MD
and PFC--NAc disconnections significantly increased the number of trials
to criterion during the set shift (see Results), we included a third control
group. This consisted of 6 rats that received unilateral inactivations of
both the PFC and either the MD or the NAc in the same hemisphere.
Results
Saline and Unilateral Inactivation Control Groups
There were no differences in the number of trails to reach criterion
during the set shift on Day 2 between rats receiving saline infusion
combinations into the PFC--MD, PFC--NAc, or MD--NAc, so their data
were combined to form one control group (F2,9 = 0.99, NS, n = 12). A
mixed ANOVA was performed on the total trials to reach criterion
during the strategy set shift for rats with unilateral inactivations, with
the brain region (NAc, PFC or MD) and side of infusion as 2 betweensubjects factor. Neither were there differences between groups on the
number of trials to reach criterion for rats receiving unilateral
inactivations (PFC n = 6, NAc n = 6, MD n = 6 [F2,12 = 0.35, NS]) nor
was there a main effect of side of infusion (F1,12 = 1.76, NS) nor a side 3
brain region interaction (F2,12 = 0.82, NS) so the data from all rats
receiving unilateral inactivations were combined to form a second
control group (Fig. 5B, inset). Lastly, there were no differences in
performance between rats that received ipsilateral inactivations of the
PFC and MD (n = 3) or PFC and NAc (n = 3; F1,4 = 0.15, NS), so their data
were combined to form the third control group.
Day 1: Response Discrimination
The results for the acquisition of the response strategy on Day 1
following mock infusions are shown in Figure 5A. A 1-way ANOVA
Table 2
Summary of treatment groups used in Experiment 2
Cannulation
n
Treatment
Group
MD--PFC
9
3
3
4
3
7
3
3
4
3
8
3
3
4
MD--bupi, PFC--bupi
MD--saline, PFC--bupi
MD--bupi, PFC--saline
MD--saline, PFC--saline
MD--bupi, PFC--bupi
MD--bupi, PFC--bupi
PFC--saline, NAc--bac/mus
PFC--bupi, NAc--saline
PFC--saline, NAc--saline
PFC--bupi, NAc--bac/mus
MD--bupi, NAc--bac/mus
MD--saline, NAc--bac/mus
MD--bupi, NAc--saline
MD--saline, NAc--saline
Disconnection
Unilateral PFC
Unilateral MD
Saline
Ipsilateral inactivations
Disconnection
Unilateral NAc
Unilateral PFC
Saline
Ipsilateral inactivations
Disconnection
Unilateral NAc
Unilateral MD
Saline
PFC--NAc
MD--NAc
Note: bupi, bupivacaine; bac/mus, baclofen/muscimol.
showed no significant difference between groups in total trials to reach
criterion for the response discrimination (F5,54 = 0.53, NS). Additionally,
there were no differences between groups on the number of probe
trials required to reach the criterion or the number of trials completed
per minute (all F values < 1.45, NS). Thus, all rats in all groups required
an equivalent number of trials to learn a response discrimination.
Day 2: Shift to Visual Cue
The results for the set shift to a visual cue--based strategy are shown in
Figure 5B. The analysis revealed a significant main effect of treatment
(F5,54 = 3.82, P < 0.005), as well as a main effect of choice (F1,54 = 269.81,
P < 0.001), and no significant treatment 3 choice interaction (F5,54 =
1.20, NS). Dunnett’s test indicated that rats receiving MD--PFC (n = 9)
and PFC--NAc (n = 7) disconnections before the set shift on Day 2
required significantly more trials to reach criterion than rats receiving
saline (both P < 0.001). Although disconnection of the MD and NAc also
increased the number of trials required to reach criterion, this increase
did not achieve statistical significance relative to saline (P = 0.14; n = 8).
Importantly, neither unilateral inactivations of each region nor combined ipsilateral inactivations of the PFC--MD or PFC--NAc increased the
number of trials to reach criterion, as these control groups did not differ
from saline-treated animals.
A mixed ANOVA on the errors committed during the set shift, with
group as the between-subjects variable, and the 3 types of errors
as levels of the within-subjects factor revealed a main effect of group
(F5,54 = 3.73, P < 0.005) and a main effect of error (F2,108 = 5.59, P <
0.001) and most pertinently, a significant group 3 error type interaction
(F10,108 = 4.87, P < 0.001). Subsequent simple main effects analyses and
Dunnett’s test revealed that rats receiving MD--PFC disconnections
made significantly more perseverative errors than saline-treated controls (P < 0.05). Similarly, the rats that received PFC--NAc disconnections also made significantly more perseverative errors (P < 0.001)
relative to saline-treated controls (Fig. 6A). Neither the unilateral or
ipsilateral inactivation control groups nor the MD--NAc disconnection
group differed from saline-treated controls in total perseverative errors.
Furthermore, there were no differences between any groups on the
number of regressive errors committed during the set shift (F5,54 =
0.61, NS, Fig. 6B). In contrast with these findings, analysis of neverreinforced errors revealed that rats receiving MD--NAc disconnection
made significantly more never-reinforced errors, relative to saline
controls (P < 0.001; Fig. 6C); all other groups did not differ from
saline-treated rats.
Additional analyses showed an effect of group for the number of trials
completed per minute during the strategy set shift (F5,54 = 4.91, P <
0.01). Multiple comparisons revealed that rats in the ipsilateral inactivation control group were actually faster to complete trials than
saline-treated rats, but no other groups differed from the saline control
group (see Table 1). Analysis of the number of probe trials required to
reach criterion showed no differences between groups (F5,54 = 0.21,
NS). Thus, disconnections between the PFC and MD, PFC and NAc, and
the MD and NAc all disrupted shifting from a response to a visual cue-based strategy, but in a dissociable manner. Disconnections between the
PFC and either the MD or the NAc caused a selective increase in
perseverative responding, whereas similar manipulations between the
MD and NAc selectively increased never-reinforced errors.
Discussion
The present data delineate a role for the MD in shifting between
behavioral strategies and demonstrate that a distributed neural
network incorporating the PFC, MD, and NAc mediate dissociable components of set shifting. Disconnecting the serial flow
of information from the MD to the PFC or the PFC projection to
the NAc resulted in a perseverative deficit, impairing the ability
of rats to inhibit the use of a previously relevant strategy or
signal the need for a shift. However, disconnections between
the MD to the NAc resulted in a unique deficit, where animals
did not perseverate, but instead displayed a selective increase
in never-reinforced errors, indicating a failure to eliminate
Cerebral Cortex July 2007, V 17 N 7 1631
Figure 5. Experiment 2: The effects of disconnection of the MD--PFC, MD--NAc, and
PFC--NAc pathways on shifting from a response to a visual cue--based strategy. Data
are expressed as means + standard error. (A) Trials to criterion on acquisition of
a response discrimination on Day 1 by rats receiving mock infusions. As there were no
differences between animals receiving unilateral (Uni: MD, NAc, or PFC, inset)
infusions on Day 2, they were combined to form one control group for analysis. (B)
Trials to criterion on the shift to the visual cue discrimination on Day 2 following
infusions of saline (white bar), unilateral inactivations (Uni, gray bars), ipsilateral
inactivations (Ipsi, striped bar), or disconnections of the MD--PFC (hatched bar), PFC-NAc (black bar), or MD--NAc (cross-hatched bar). Neither saline infusions nor unilateral
inactivations impaired rats’ ability to shift strategy. However, rats receiving
disconnections of the MD--PFC and PFC--NAc pathways showed significant increases
in total trials to reach criterion relative to saline control group (double stars denote P <
0.01 vs. saline control group). (C) Schematic representing the 3 types of asymmetrical
disconnection treatments used in Experiment 2.
inappropriate strategies during the set shift. Notably, neither
unilateral inactivation of any of the targeted structures nor
ipsilateral inactivations impaired behavior, further supporting
the contention that the impairments were due specifically to
functional disconnection and not due to additive effects of
unilateral inactivations.
1632 Thalamic--Prefrontal Circuits and Set Shifting
d
Block and others
Bilateral MD Inactivations
Bilateral inactivation of the MD impaired shifting between
response and visual cue--based discrimination strategies, without affecting initial discrimination learning, indicating that the
MD is specifically involved in shifting between strategies. These
impairments were due to an increase in the number of perseverative errors, which implies a disruption in the ability to disengage from a previously relevant strategy. In addition, these
treatments also increased the number of never-reinforced
errors, although this effect only achieved statistical significance
on the more difficult visual cue to response shift. This latter
finding indicates that the MD also may play a role in eliminating
inappropriate response alternatives based on negative feedback,
particularly in situations when animals are required to perform
more difficult shifts.
The strategy set-shifting task used in the present study
required rats to cease attending to a previously relevant
stimulus dimension (e.g., turn direction) and attend to a newly
relevant stimulus (e.g., the visual cue) in order to obtain reward.
However, given that the same stimuli were used during the
initial discrimination and during the set shift, it may be argued
that this task also assesses a form of reversal learning. In this
regard, a number of studies have assessed the effects of MD
lesions on different types of reversals. Tigner (1974) reported
that MD lesions induced deficits in both the initial learning
and reversal of brightness and tactile discriminations, although
no effect was seen for spatial reversals. Using a more rule-based
reversal in the operant chamber, Beracochea and others (1989)
failed to find any effect of MD lesion on reversals. On the other
hand, MD lesions did not impair learning of an initial spatial
reversal but did disrupt subsequent reversals on a T-maze task
(Means and others 1975), whereas other studies failed to find
any effect on serial spatial reversals performed in an operant
chamber (Neave and others 1993). With respect to reversal
learning for visual stimuli, Chudasama and others (2001)
demonstrated that excitotoxic lesions of the MD did impair
reversals of visually cued stimulus--reward associations but
that this effect only emerged on latter serial reversals while
having a minimal effect on the initial reversal. Taken together,
it is apparent that the MD may play a more prominent role in
serial reversal learning for visual or spatial discriminations but
may not be as important for initial reversal learning. Given
that rats in the present study were only required to shift
strategies once, it is unlikely that the impairments induced by
MD inactivation reported here were due to a disruption of
processes related to reversal learning.
Although the role of the MD in reversal learning is unclear,
lesions of this nucleus do cause robust perseverative deficits
when rats are required to shift from a matching to a nonmatching to place strategy on a T-maze task (Hunt and Aggleton
1998), indicating that this form of behavioral flexibility is
critically dependent on the integrity of the MD. A similar
perseverative deficit was observed in the present study, and
this type of impairment is similar to what has been reported
following medial PFC manipulations (Ragozzino and others
1999; Ragozzino 2002; Floresco, Magyar, and others 2006).
This finding was not unanticipated, considering the strong
reciprocal connections between the MD and the medial PFC
(Condé and others 1995; Gabbott and others 2005). A number
of studies have demonstrated that lesions or inactivations of
the MD can induce similar behavioral deficits to those observed
Figure 6. Analysis of the type of errors committed during the set shift in Experiment 2. (A) MD--PFC (hatched bar) and PFC--NAc (black bar) disconnections resulted in a significant
increase in perseverative errors, whereas unilateral inactivations (Uni, gray bar), ipsilateral inactivations (Ipsi, striped bar), and MD--NAc disconnections (cross-hatched bar) were
without effect compared with saline treatments (white bar; single and double stars denote P < 0.005, 0.01, vs. saline control group, respectively). (B) None of the disconnection
treatments or unilateral/ipsilateral inactivations altered the number of regressive errors. (C) Only MD--NAc disconnections resulted in an increase in never-reinforced type errors
relative to saline controls (double stars denote P < 0.01 vs. saline control group).
following PFC manipulations. These include different types of
delayed response tasks in both primates and rats (Harrison and
Mair 1996; Floresco and others 1999; de Zubicaray and others
2001; Funahashi and others 2004; Bailey and Mair 2005) and
familiarity estimates in humans (Zoppelt and others 2003). The
present findings indicate that the MD also plays a key role in
set shifting and indicate that this region may interact with the
PFC to facilitate this form of behavioral flexibility; a hypothesis
tested directly in Experiment 2.
MD--PFC Disconnections
Asymmetrical bupivacaine infusions into the MD and PFC,
which disconnected the flow of information between these
structures, increased the total trials to reach criterion on a
visual cue discrimination during the strategy set shift. This
deficit was due to a selective increase in perseverative errors,
indicating that this thalamocortical pathway is necessary for
successful disengagement from a previously relevant strategy.
The prelimbic region of the medial PFC has been shown to have
a role in maintaining an established outcome--action association
in working memory for the purpose of guiding future behavior
(Corbit and Balleine 2003). In that study, rats with lesions to the
prelimbic PFC were trained to differentially respond for sucrose solution and sugar pellets, one of which was subsequently
experimentally devalued with a specific satiety manipulation.
Lesioned rats failed to differentiate between the devalued and
nondevalued rewards and globally depressed responding. These
results indicate that the PFC is crucial for the incorporation
of specific reward information into behavior (Corbit and
Balleine 2003). Furthermore, lesions of the MD reveal a similar
deficit (Corbit and others 2003). When these findings are viewed
in light of the present data, they suggest that the MD and the
PFC work in concert to suppress the use of previously established rules based on irrelevant reinforcement contingencies;
a contention supported by the present findings that disconnection between the MD and PFC impaired the suppression of
a previously relevant strategy during the set shift.
Further insight into the specific roles that the MD and PFC
may play in mediating behavioral flexibility comes from functional imaging studies in humans. Thalamic activation has been
linked to performance of the WCST, where activation of the MD
occurred specifically when the participants received negative feedback after an incorrect choice but not positive
feedback (Monchi and others 2001). These findings suggest
that increased activity in the MD serves as a signal for the
necessity to shift set. In contrast, activation of the dorsolateral
PFC occurred during both positive and negative feedback, indicating the PFC is critically involved in comparing online feedback with past events (Monchi and others 2001). Along similar
lines, MD neurons recorded from awake, behaving rabbits
learning a discriminative avoidance task display earlier and
more robust development of discriminative activity during
reversal of stimulus-reinforcement contingencies when compared with similar activity of PFC neurons (Orona and Gabriel
1983). Thus, changes in reinforcement contingencies are
associated with an initial alteration in the discriminative activity
of MD neurons, which subsequently can modulate PFC neural
activity to facilitate changes in response strategies. As such,
even though inactivation of either the PFC (Ragozzino and
others 1999) or the MD (present study) caused a similar
perseverative deficit during the set shift, each of these regions
may mediate different cognitive operations that facilitate
shifting from one strategy to another. It follows that disconnection of the MD and PFC would deprive the PFC of
information related to the absence of reward (mediated initially
by the MD) after entering an incorrect arm during the set shift.
This in turn would impair the ability to extinguish responding in
accordance with the previously relevant strategy so that a new
strategy might be initiated (Lebron and others 2004).
PFC--NAc Disconnections
As was observed following MD--PFC disconnections, asymmetrical inactivations of the PFC and the NAc also impaired the
ability to shift from one discrimination strategy to another,
causing a selective increase in perseverative errors. However,
these manipulations did not increase the number of regressive
or never-reinforced errors, which are used as an index of an
animal’s ability to maintain and establish a new strategy during
the shift (Ragozzino 2002; Ragozzino and others 2002; Floresco,
Ghods-Sharifi, and others 2006; Floresco, Magyar, and others
2006). This pattern of errors is interesting, in light of the effect
of bilateral inactivations of the NAc on this task (Floresco,
Ghods-Sharifi, and others 2006). Specifically, inactivation of
the NAc increased regressive and never-reinforced errors but
did not affect perseverative responding. We have interpreted
these findings to indicate that the NAc is not involved in the
suppression of previously acquired strategies, but instead facilitates the maintenance of a new strategy, potentially working
in conjunction with the dorsomedial striatum (Ragozzino and
other 2002; Floresco, Ghods-Sharifi, and others 2006).
Cerebral Cortex July 2007, V 17 N 7 1633
Although both perseverative and regressive errors entail
a choice consistent with the previously acquired, but now
incorrect, strategy, a key difference is when these errors occur
in the choice sequence. Regressive errors occur later in the
choice sequence when rats use the previous strategy on fewer
than 50% of trials where the correct response is opposite to that
required using the initial rule. Thus, perseverative and regressive errors are inexorably related as an animal cannot
maintain a novel strategy unless it has first shifted from the
previously rewarded strategy. We have previously proposed
that the PFC and NAc play distinct roles during different stages
of set shifting. Specifically, PFC plays an essential role early in set
shifting, suppressing the use of a previously correct strategy
(Ragozzino and others 1999), whereas once the appropriate
new strategy has been ascertained, maintenance of the novel
strategy is likely mediated by parallel corticostriatal circuits that
include NAc (Floresco, Ghods-Sharifi, and others 2006). The
finding that bilateral inactivation of the NAc increased regressive but not perseverative errors (Floresco, Ghods-Sharifi,
and others 2006), whereas PFC--NAc disconnection increased
perseverative responding suggests that this corticostriatal
circuit facilitates both the suppression of previously established
response rules and the maintenance of novel strategies. The
increase in perseverative errors following PFC--NAc disconnections may be attributed to disruption of the transfer of updated
information from the PFC regarding the success or failure of
current strategies to the NAc for maintenance of a novel discrimination strategy. This hypothesis is consistent with previous studies investigating the role of this pathway in the
context of a 5-choice serial reaction time task (Christakou and
others 2004). In that study, disconnection of the PFC--NAc
pathway caused a deficit in the integration of reward information into the behavioral repertoire. Likewise, asymmetrical inactivations between the PFC and the NAc also impaired working
memory performance on a delayed response version of the
radial arm maze task (Floresco and others 1999). Taken
together, these data support the contention that interactions
between the PFC and NAc play a crucial role in the maintenance
of behavioral strategies in response to changes in reward
contingencies, as well as updating responding based on trialunique reward information. These include sustained attention
(Christakou and others 2004), working memory (Floresco and
others 1999), and, as demonstrated here, set shifting.
MD--NAc Disconnections
Disconnection inactivations of the MD to the NAc resulted
in a unique pattern of errors during the set shift, distinct from
that induced by either the MD--PFC or PFC--NAc disconnections.
Specifically, we observed a selective increase in never-reinforced
type errors, with no concomitant increase in perseverative or
regressive errors. As opposed to perseverative/regressive errors,
never-reinforced errors are scored when an animal makes
choice that was incorrect during both initial discrimination
training and during the shift and are used as an index of how
quickly animals are able to parse out ineffective strategies.
During the set shift, intact rats learn quickly that the previously
correct strategy is no longer appropriate and engage alternative
strategies until they find the optimal solution. Never-reinforced
errors may be interpreted as an attempt to use alternative
strategies, perhaps a reversal of the previously acquired rule
(e.g., always turn left instead of right). Intact rats make relatively
few errors of this type and learn that these strategies do not lead
1634 Thalamic--Prefrontal Circuits and Set Shifting
d
Block and others
to reward reliably. In contrast, bilateral inactivation of the NAc
caused a disproportionate amount of never-reinforced errors
(as well as regressive errors), indicating that the NAc also
mediates the elimination of inappropriate response options, enabling the reorganization of behavior to obtain reward in an
optimal manner (Floresco, Ghods-Sharifi, and others 2006). From
the present data, it is apparent that this thalamostriatal pathway plays a specialized role in establishing novel strategies,
facilitating the elimination of ineffective response options when
reward contingencies have changed. To our knowledge, this is
the first characterization of a functional role for this pathway. In
contrast, disconnections of this pathway did not impair working
memory assessed using a delayed response variant of the radial
arm maze (Floresco and others 1999), indicating that this pathway does not mediate working memory functions subserved
by the PFC. Yet, the increase in never-reinforced errors observed
in the present study indicates information transfer from the
MD to the NAc is necessary for the successful elimination of
inappropriate strategies based on reward feedback. Together,
these findings highlight the differences in the functional neural
circuitries that underlie different executive functions (i.e.,
working memory and behavioral flexibility) mediated by the
frontal lobes.
When viewed collectively, the data from Experiment 2
further clarify the functional roles for parallel thalamo-corticostriatal circuits involved in the overall process of strategy set
shifting and provide important information about the functions
of specific subcircuits that mediate dissociable components of
behavioral flexibility. Based on the present data, in addition to
previous findings, we propose the following. Early in the set
shift, increased activity in the MD encodes information about
changes in reward contingencies and subsequently relays this
information to the PFC (Fig. 7A). In turn, the PFC serves to
suppress the use of a previously correct, but now irrelevant,
strategy, a process facilitated by dopaminergic inputs to this
region (Ragozzino and others 1999; Ragozzino 2002; Floresco,
Magyar, and others 2006). Once perseveration has ceased and
the animal engages new strategies, the activity in the MD--NAc
pathway facilitates new learning by curtailing the use of inappropriate strategies (Fig. 7B). Subsequently, when the appropriate strategy has been ascertained, maintenance of the novel
Figure 7. Schematic of some of the neural circuitry that mediates strategy set
shifting. See Discussion for details.
strategy is mediated by corticostriatal circuits that include the
PFC and NAc (Fig. 7C).
Conclusions
Behavioral flexibility is a composite of different processes:
suppression of irrelevant strategies, acquisition and generation
of novel strategies, and maintenance of effective strategies. The
present data emphasize that shifting from one behavioral
strategy to another is mediated by a distributed neural circuit
and clarify the contributions that different cortical subcortical
regions make to this form of behavioral flexibility. Our findings
that dissociable components of different phases of strategy set
shifting are mediated by distinct neural circuits complement
the findings of human functional imaging studies, with different
patterns of activation of the PFC, thalamus, or striatum
associated with different cognitive requirements of the WCST
(Monchi and others 2001). Here, we establish that parallel
pathways originating from the MD and terminating in both the
PFC and NAc make distinct contributions to behavior when an
organism must change its behavior in response to dynamic
environmental demands. A functional projection from the MD
to the PFC is important for signaling the need to shift strategies,
and the projection to the NAc is important for eliminating
inappropriate response strategies. Furthermore, the PFC projection to the NAc facilitates the integration of reward information and behavioral consequences into behavioral
responses geared toward the anticipation of future reward.
It is interesting to note that there is evidence of compromised
MD--PFC circuitry in schizophrenia, as this disorder is associated
with reductions in the number of neurons and overall volume of
the MD (Andreasen and others 1994; Young and others 2000;
Dorph-Petersen and others 2004). Additionally, schizophrenic
individuals show deficits in tasks assessing behavioral flexibility
that are rooted in impaired concept formation (Pantelis and
others 1999) and often display perseverative responding in
attentional set-shifting paradigms (Elliott and others 1998).
When viewed in light of the present findings, it is possible
that the deficits in behavioral flexibility observed in schizophrenia may be due in part to a compromised projection originating
in the MD and terminating in the PFC and NAc. Collectively,
these findings further elucidate the neural circuitry that
regulates behavioral flexibility in the normal brain and provide
important insight into the neural pathology that underlies
executive dysfunction in some psychiatric disorders where
impairment in behavioral flexibility is a prominent symptom.
Notes
This work was supported by Discovery Grant from the Natural Sciences
and Engineering Research Council of Canada to SBF. SBF is a Canadian
Institutes of Health Research New Investigator, a Michael Smith Foundation for Health Research Scholar, and the recipient of a National Alliance
for Research on Schizophrenia and Depression Young Investigator Award.
We are indebted to Sarvin Ghods-Sharifi and Desirae Haluk for their
assistance with behavioral testing. Conflict of Interest: None declared.
Address correspondence to Stan B. Floresco, Department of Psychology and Brain Research Center, University of British Columbia, 2136
West Mall, Vancouver, British Columbia, Canada V6T 1Z4. Email:
fl[email protected].
References
Andreasen NC, Arndt S, Cizadlo T, Flaum M, O’Leary D, Ehrhardt JC, Yuh
WT. 1994. Thalamic abnormalities in schizophrenia visualized
through magnetic resonance image averaging. Science 266:294--298.
Bailey KR, Mair RG. 2005. Lesions of specific and nonspecific thalamic
nuclei affect prefrontal cortex-dependent aspects of spatial working
memory. Behav Neurosci 119:410--419.
Beracochea DJ, Jaffard R, Jarrard LE. 1989. Effects of anterior or
dorsomedial thalamic ibotenic lesions on learning and memory in
rats. Behav Neural Biol 51:364--376.
Berendse HW, Groenewegen HJ. 1990. Organization of the thalamostriatal projections in the rat, with special emphasis on the ventral
striatum. J Comp Neurol 299:187--228.
Birrell JM, Brown V. 2000. Medial frontal cortex mediates perceptual
attentional set shifting in the rat. J Neurosci 20:4320--4324.
Brog JS, Apongse AS, Deutch AY, Zahm DS. 1993. The patterns of afferent
innervation of the core and shell in the ‘‘accumbens’’ part of the rat
ventral striatum: immunohistochemical detection of retrogradely
transported fluoro-gold. J Comp Neurol 338:255--278.
Christakou A, Robbins TW, Everitt BJ. 2004. Prefrontal cortical--ventral
striatal interactions involved in affective modulation of attentional
performance: implications for corticostriatal circuit function.
J Neurosci 24:773--780.
Chudasama Y, Bussey TJ, Muir JL. 2001. Effects of selective thalamic and
prelimbic cortex lesions on two types of visual discrimination and
reversal learning. Eur J Neurosci 14:1009--1020.
Chudasama Y, Passetti F, Rhodes SE, Lopian D, Desai A, Robbins TW. 2003.
Dissociable aspects of performance on the 5-choice serial reaction
time task following lesions of the dorsal anterior cingulate, infralimbic
and orbitofrontal cortex in the rat: differential effects on selectivity,
impulsivity and compulsivity. Behav Brain Res 146:105--119.
Condé F, Maire-LePoivre E, Audinat E, Crépel F. 1995. Afferent
connections of the medial frontal cortex of the rat. II. Cortical and
subcortical afferents. J Comp Neurol 352:567--593.
Corbit LH, Balleine BW. 2003. The role of the prelimbic cortex in
instrumental conditioning. Behav Brain Res 146:145--157.
Corbit LH, Muir JL, Balleine BW. 2003. Lesions of mediodorsal thalamus
and anterior thalamic nuclei produce dissociable effects on instrumental conditioning in rats. Eur J Neurosci 18:1286--1294.
de Zubicaray GI, McMahon K, Wilson SJ, Muthiah S. 2001. Brain activity
during the encoding retention and retrieval of stimulus representations. Learn Mem 8:243--251.
Dias R, Aggleton JP. 2000. Effects of selective excitotoxic prefrontal
lesions on acquisition of nonmatching- and matching-to-place in
the T-maze in the rat: differential involvement of the prelimbicinfralimbic and anterior cingulate cortices in providing behavioral
flexibility. Eur J Neurosci 12:4457--4466.
Dias R, Robbins TW, Roberts AC. 1997. Dissociable forms of inhibitory
control within prefrontal cortex with an analog of the Wisconsin
card sort test: restriction to novel situations and independence from
‘‘on-line’’ processing. J Neurosci 17:9285--9297.
Dorph-Petersen K-A, Pierri JN, Sun Z, Sampson AR, Lewis DA. 2004. Stereological analysis of the mediodorsal thalamic nucleus in schizophrenia:
volume, neuron number, and cell types. J Comp Neurol 472:449--462.
Dunnett SB, Meldrum A, Muir JL. 2005. Frontal-striatal disconnection
disrupts cognitive performance of the frontal-type in the rat.
Neuroscience 135:1055--1065.
Elliott R, McKenna PJ, Robbins TW, Sahakian BJ. 1998. Specific neuropsychological deficits in schizophrenic patients with preserved
intellectual function. Cogn Neuropsychiatry 3:45--70.
Floresco SB, Braaksma DN, Phillips AG. 1999. Thalamic-cortical-striatal
circuitry subserves working memory during delayed responding on
a radial arm maze. J Neurosci 19:11061--11071.
Floresco SB, Ghods-Sharifi S. 2006. Amygdala-prefrontal cortical circuitry regulates effort-based decision making. Cereb Cortex
10.1093/cercor/bhj143.
Floresco SB, Ghods-Sharifi S, Vexelman C, Tse MT. 2006. Dissociable
roles for the nucleus accumbens core and shell in regulating setshifting. J Neurosci 26:2449--2457.
Floresco SB, Magyar O, Ghods-Sharifi S, Vexelman C, Tse MT. 2006.
Multiple dopamine receptor subtypes in the medial prefrontal cortex
of the rat regulate set-shifting. Neuropsychopharmacology 31:297--309.
Floresco SB, Seamans JK, Phillips AG. 1997. Selective roles for hippocampal, prefrontal cortical, and ventral striatal circuits in radial-arm
maze tasks with or without a delay. J Neurosci 17:1880--1890.
Cerebral Cortex July 2007, V 17 N 7 1635
Funahashi S, Takeda K, Watanabe Y. 2004. Neural mechanisms of spatial
working memory: contributions of the dorsolateral prefrontal cortex
and the thalamic mediodorsal nucleus. Cogn Affect Behav Neurosci
4:409--420.
Gabbott PL, Warner TA, Jays PR, Salway P, Busby SJ. 2005. Prefrontal
cortex in the rat: projections to subcortical autonomic, motor, and
limbic centers. J Comp Neurol 492:145--177.
Gauntlett-Gilbert J, Roberts RC, Brown VJ. 1999. Mechanisms underlying
attentional set-shifting in Parkinson’s disease. Neuropsychologia
37:605--616.
Gorelova N, Yang CR. 1997. The course of neural projection from the
prefrontal cortex to the nucleus accumbens in the rat. Neuroscience
76:689--706.
Groenewegen HJ. 1988. Organization of the afferent connections of the
mediodorsal thalamic nucleus in the rat, related to the mediodorsalprefrontal topography. Neuroscience 24:379--431.
Harrison LM, Mair RG. 1996. A comparison of the effects of frontal
cortical and thalamic lesions on measures of spatial learning and
memory in the rat. Behav Brain Res 75:195--206.
Hunt PR, Aggleton JP. 1998. Neurotoxic lesions of the dorsomedial
thalamus impair the acquisition but not the performance of delayed
matching to place by rats: a deficit in shifting response rules.
J Neurosci 18:10045--10052.
Joyce EM. 1987. The neurochemistry of Korsakoff’s syndrome. In: Stahl
SM, Iversen SD, Goodman EC, editors. Cognitive neurochemistry.
Oxford: Oxford University Press. p 300--337.
Kim J, Ragozzino ME. 2005. The involvement of the orbitofrontal cortex
in learning under changing task contingencies. Neurobiol Learn
Mem 83:125--133.
Lawrence AD, Sahakian BJ, Rogers RD, Hodge JR, Robbins TW. 1999.
Discrimination, reversal, and shift learning in Huntington’s disease:
mechanisms of impaired response selection. Neuropsychologia
37:1359--1374.
Lebron K, Milad MR, Quirk GJ. 2004. Delayed recall of fear extinction in
rats with lesions of ventral medial prefrontal cortex. Learn Mem
11:544--548.
Lombardi WJ, Andreason PJ, Sirocco KY, Rio DE, Gross RE, Umhau JC,
Hommer DW. 1999. Wisconsin card sorting test performance
following head injury: dorsolateral fronto-striatal circuit activity
predicts perseveration. J Clin Exp Neuropsychol 21:2--16.
McAlonan K, Brown VJ. 2003. Orbital prefrontal cortex mediates
reversal learning and not attentional set shifting in the rat. Behav
Brain Res 146:97--103.
McFarland K, Kalivas PW. 2001. The circuitry mediating cocaineinduced reinstatement of drug-seeking behavior. J Neurosci
21:8655--8663.
Means LW, Hershey AE, Waterhouse GJ, Lane CJ. 1975. Effects of
dorsomedial thalamic lesions on spatial discrimination reversal in the
rat. Physiol Behav 14:725--729.
Monchi O, Petrides M, Petre V, Worsley K, Dagher A. 2001. Wisconsin
card sorting revisited: distinct neural circuits participating in
different stages of the task identified by event-related functional
magnetic resonance imaging. J Neurosci 21:7733--7741.
1636 Thalamic--Prefrontal Circuits and Set Shifting
d
Block and others
Neave N, Sahgal A, Aggleton J. 1993. Lack of effect of dorsomedial
thalamic lesions on automated tests of spatial memory in the rat.
Behav Brain Res 55:39--49.
Olton DS, Walker JA, Wolf WA. 1982. A disconnection analysis of
hippocampal function. Brain Res 233:241--253.
Orona E, Gabriel M. 1983. Multiple-unit activity of the prefrontal cortex
and mediodorsal thalamic nucleus during reversal learning of
discriminative avoidance behavior in rabbits. Brain Res 263:313--329.
Oscar-Berman M, Kirkley SM, Gansler DA, Couture A. 2005. Comparisons
of Korsakoff and non-Korsakoff alcoholics on neuropsychological
tests of prefrontal brain functioning. Alcohol Clin Exp Res 20:667--675.
Owen AM, Roberts AC, Hodges JR, Summers BA, Polkey CE, Robbins TW.
1993. Contrasting mechanisms of impaired attentional set-shifting in
patients with frontal lobe damage or Parkinson’s disease. Brain
116:1159--1175.
Pantelis C, Barber FZ, Barnes TR, Nelson HE, Owen AM, Robbins TW.
1999. Comparison of set-shifting ability in patients with chronic
schizophrenia and frontal lobe damage. Schizophr Res 37:251--270.
Paxinos G, Watson C. 1998. The rat brain in stereotaxic coordinates. 4th
ed. San Diego, CA: Academic Press.
Popken GJ, Bunney WE Jr, Potkin SG, Jones EG. 2000. Subnucleusspecific loss of neurons in medial thalamus of schizophrenics. Proc
Natl Acad Sci USA 97:9276--9280.
Ragozzino ME. 2002. The effects of dopamine D1 receptor blockade on
the prelimbic-infralimbic areas on behavioral flexibility. Learn Mem
9:18--28.
Ragozzino ME, Detrich S, Kesner RP. 1999. Involvement of the
prelimbic-infralimbic areas of the rodent prefrontal cortex in
behavioral flexibility for place and response learning. J Neurosci
19:4585--4594.
Ragozzino ME, Ragozzino KE, Mizumori SJ, Kesner RP. 2002. Role of the
dorsomedial striatum in behavioral flexibility for response and visual
cue discrimination learning. Behav Neurosci 116:105--115.
Stefani MR, Groth K, Moghaddam B. 2003. Glutamate receptors in the rat
medial prefrontal cortex regulate set-shifting ability. Behav Neurosci
117:728--737.
Stuss DT, Levine B, Alexander MP, Hong J, Palumbo C, Hamer L, Murphy
KJ, Izukawa D. 2000. Wisconsin card sorting test performance in
patients with focal frontal and posterior brain damage: effects of
lesion location and test structure on separable cognitive processes.
Neuropsychologia 38:388--402.
Tigner JC. 1974. The effects of dorsomedial thalamic lesions on learning,
reversal, and alternation behavior in the rat. Physiol Behav 12:13--17.
Warburton EC, Baird AL, Morgan A, Muir JL, Aggleton JP. 2000.
Disconnecting hippocampal projections to the anterior thalamus
produces deficits on tests of spatial memory in rats. Eur J Neurosci
12:1714--1726.
Young KA, Manaye KF, Liang C-L, Hicks PB, German DC. 2000. Reduced
number of mediodorsal and anterior thalamic neurons in schizophrenia. Biol Psychiatry 47:944--953.
Zoppelt D, Koch B, Schwarz M, Daum I. 2003. Involvement of the
mediodorsal thalamic nucleus in mediating recollection and familiarity. Neuropsychologia 41:1160--1170.