Thalamic lesion effects on aversive behaviors


Material Information

Thalamic lesion effects on aversive behaviors
Physical Description:
viii, 164 leaves : ill. ; 29 cm.
Mauderli, Andre P., 1947-
Publication Date:


Subjects / Keywords:
Research   ( mesh )
Thalamus -- physiology   ( mesh )
Photic Stimulation   ( mesh )
Physical Stimulation   ( mesh )
Stereotaxic Techniques   ( mesh )
Pain Measurement   ( mesh )
Narcotics -- pharmacology   ( mesh )
Escape Reaction   ( mesh )
Rats   ( mesh )
Department of Neuroscience thesis Ph.D   ( mesh )
Dissertations, Academic -- College of Medicine -- Department of Neuroscience -- UF   ( mesh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph.D.)--University of Florida, 1996.
Bibliography: leaves 152-162.
Statement of Responsibility:
by Andre P. Mauderli.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002298750
oclc - 48924885
notis - ALQ2012
System ID:

This item is only available as the following downloads:

Full Text








I would not have been able to work and study at this
institution without Harry Lundeen, who made it possible for me to
immigrate to America. I will always be grateful to Harry and his
wife Lyla for what they did for me. I would like to thank Parker
Mahan, who has introduced me to the field of facial pain medicine
and who is my role model because he cares for his patients and
for his colleagues. I would like to express my appreciation to
Arnold Bleiweis who has always encouraged me and provided
generous support through a NRSA traineeship. I am indebted to my
mentors Charles Vierck, Louis Ritz and Richard Fessler for their
valuable advice, for letting me work in their laboratories and for
providing equipment and supplies. I was fortunate to have Charles
Vierck as my primary advisor. He gave me the freedom to develop
my own project, helped me to submit a successful application for
an Individual Dentist Scientist Award and was always accessible
for discussions that taught me more than many courses. I
appreciate the support I received from Bill Luttge, who allowed
me to join the Neuroscience Graduate Program and gave me
encouragement whenever I needed it. Paul Kubilis contributed his
expertise as a statistician and went out of his way to get this
project completed on time. I greatly appreciate his help. I am
grateful to Jaekyeong Baek for her encouragement, support and

valuable advice and I am thanking my mother back in Switzerland
for her understanding and love throughout all these years.
This research was supported through the Department of Oral
Biology (PI: Dr. Arnold S. Bleiweis) by the NRSA training grant T32
DE07200 and by the Individual Dentist Scientist Award K15


ACKNOWLEDGMENTS........................ ................ ............................. i i

ABSTRACT .......................................................................................... vii

INTRODUCTION............................... ................. ................................ 1

NOCICEPTION...................................................................................... 4

Introduction ................................................................................ 4
Design Criteria for New Paradigm ...................................... 8
Description of New Paradigm ............................................... 9
Equipment for Escapetest............................ ......... 10
Test Protocol for Escapetest .................................... 12
Response Variables for Escapetest........................ 12
Equipment for Darkboxtest ...................................... 13
Test Protocol for Darkboxtest................................. 15
Response Variables for Darkboxtest................... 15
The Darkboxtest as a Control Test ....................... 15
Data Collection Method................................. ..... 17
Statistical Analysis of Individual Animal Data 18
Validation Experiments ...................................... ....... 22
Animal Subjects ........................................ ....... 22
Training of the Subjects ................................ ...... 23
Validation of the Escapetest.................................. 23
Validation of the Darkboxtest................................ 37
Conclusion ........................................................... ................ 44


Introduction ......................................................... ............... 45
Stereotaxic Methodology................................................. 46
Coordinate System.............................................. 46

Stereotaxic Frame .................................... .......... 49
Stereotaxic Errors .................................... .......... 49
Computer Aided Methodology.................................... 61
Lesion Methodology.................................................................. 63
Radiofrequency Lesions ............................................. 64
Conclusion ...................................... .............. ....................... 68

BEHAVIORS OF THE RAT .................................................. .......... 70

Introduction .............................................................................. 70
Anatomical and Physiological Background.................. 71
General Topographic Anatomy.................................. 71
Anatomy and Physiology of Peripheral Inputs... 72
Central Fiber Connections ....................................... 75
Conclusion ....................................................................... 78
Behavioral Data ..................................... ............... 79
Introduction ............................................................... 79
Lesion Effect on Nociception and Opioid Action 79
Lesion Size ........................................ ........... ............. 80
Lesion Effect on Memory ........................................... 82
Role of Test Structure ................................ ...... 83
Clinical Data............................................ .............................. 86
Sum m ary........................................................ ................. 89
Research Goals ..................................... ................ 90
Specific Aims ..................................... ...... 92
Hypotheses .................................................... ..... ........ 93
Organization of the Study ......... ...................................... 95
Materials and Methods............................. ........... 96
Animal Subjects ................................ ........ 96
Behavioral Tests.......... ........ ........ 97
Test Schedule ............................................................. .. 99
Statistical A analysis ..................................................... 104
Radiofrequency Lesioning ......................................... 106
Histology................................................... .. ................ 108
Results..................................... ................................................ 112
Histological Findings .... ........ ........... 112
Behavioral Results of Lateral vs. Medial
Lesion Experiments ....................................................... 114
Behavioral Results of the Lesion/Morphine
Experim ent ..................................................... .. .......... 132
Discussion ................................ ....................................... ..... .. 143

CONCLUSION ............................................................................................. 149

REFERENCES........................................................................................ 152

BIOGRA PHICA L SKETCH ........................................................................ 163

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment
of the Requirements for the Degree of
Doctor of Philosophy


Andre P. Mauderli
August 1996

Chairman: Charles J. Vierck
Major Department: Neuroscience

Intractable pain has a devastating effect on the quality of
life of patients and represents a great economic burden for
society. The development of effective treatments for some pain
conditions depends on a better understanding of underlying
mechanisms. The thalamus has long been known to play a role in
pain processing, and its medial part is suspected to be involved in
affective motivational aspects of pain. This study investigated
the parafascicular nucleus of the medial thalamus. Existing
knowledge led to the hypotheses that the role of this nucleus in
conscious nociception could be (1) affective/motivational, (2)
motor, (3) sensory, or (4) to modulate antinociceptive opioid
systems. A common method for investigating the function of
thalamic nuclei has been to measure behavioral responses to
nociceptive stimulation, before and after experimentally ablating
the nucleus of interest and to infer function from lesion-induced
behavioral changes. In past lesion studies of the parafascicular

region the interpretation of results was hampered by lesions that
were not selective to the intended target, and by behavioral
methods that could not distinguish between lesion effects on
motivation for a response or effects on execution of the response.
A first contribution of this study was the development of
refined stereotaxic methodology to enhance the accuracy of the
lesions. In combination with radiofrequency lesion techniques it
helped to minimize collateral damage to unintended structures.
A second contribution was the development of a new
behavioral assay for measuring conscious nociception. The assay
used two tests which both measured operant escape. The use of
different motivators--nociceptive heat and bright light--to elicit
the same response--escape--allowed discrimination between
sensory and motor effects.
The third contribution was an investigation of the
functional role of the parafascicular nucleus in rodents--using
the new methodology. The results were consistent with the
hypothesis that antinociceptive actions of opioids are increased
following ablation of the parafascicular nucleus.


Intractable chronic pain has a devastating effect on the
quality of life of patients and represents a great economic burden
for society. For certain pain conditions, no treatment is available
that provides lasting relief. Even neurosurgical ablation of
components of the central nervous system that mediate pain
often brings only temporary relief, and when pain returns, it may
be more severe and more therapy-resistant than the original pain.
Effective treatments for intractable pain may emerge from a
better understanding of where and how the affective dimension of
pain is generated in the brain. The interest of clinicians in
aversive, motivational qualities of pain is high because these
aspects cause suffering and interfere with a productive life. The
thalamus has long been known to play a role in pain processing,
and its medial part is suspected to be involved in affective
motivational aspects of pain. The present research focuses on a
nucleus in the medial thalamus, the parafascicular nucleus, the
functional role of which is under dispute.
A common method for investigating the function of
components of the central nervous system is to compare
behavioral responses before and after experimentally ablating the

structure of interest in animals. The interpretation of results
obtained with this approach has often been hampered by
behavioral assays that failed to distinguish between sensory and
motor lesion effects and by poor selectivity of the lesions. The
present research attempted to improve upon this situation: (1) by
introducing a new combination of behavioral tests that allowed to
discriminate between sensory and motor lesion effects by using
different aversive stimulus modalities to elicit the same type of
response; (2) by identifying factors that compromise the
stereotaxic accuracy of lesions and developing techniques to
minimize the detrimental impact of these factors; (3) by applying
the newly developed behavioral and lesion methodology in a study
of the putative aversive role of the parafascicular thalamic
The methodological contributions were intended not only for
the realization of the present study of thalamic function but for a
wider audience interested in investigating nociceptive systems
inside and outside the thalamus.
A goal of the parafascicular lesion study was to evaluate
functional hypotheses for this thalamic region. According to a
first hypothesis the parafascicular nucleus has a role in
generating the behavioral driving force for a voluntary response
such as escape. A second hypothesis defines the parafascicular
nucleus as a motor nucleus which controls the selection between
unlearned and learned response options. According to a third
hypothesis the parafascicular nucleus has a sensory role which
underlies the triggering of both unlearned and learned responses.

A fourth hypothesis proposes the parafascicular nucleus to be a
modulator of anti-nociceptive opioid systems.
The present research, by adding to existing knowledge, can
make a contribution toward clinical progress in the control of
intractable pain: it may help direct the attention toward other
thalamic candidate sites for aversive processing, if it turns out
that the parafascicular nucleus is primarily a motor nucleus. On
the other hand, if there is support for a major sensory or
modulatory role, it may be time to screen the intrinsic signaling
systems of the nucleus on a molecular level for possibilities of
selective pharmacological interventions. The present work may
help to better define the pain-relieving potential of surgical or
pharmacological interventions in the parafascicular nucleus of
human patients.



The ultimate goal of most pain research is to develop
therapeutic methods that eliminate or reduce pain in humans. An
important step toward this goal is to advance the understanding
of neural systems that mediate pain. Mechanisms that initiate,
enhance or reduce pain exist at all levels, from molecular
processes within individual cells to chemical and electrical
interactions between neurons to signaling and feedback pathways
along the neuraxis and finally to networks with nodes throughout
the central nervous system. The present project is focused on the
network level, specifically on the role of nuclei within the medial
Pain at the systems or network level has many dimensions.
During a peripheral noxious event, it may provide precise
temporal and topographic information about the insult and
represent the sensory component of protective reflexes. Repeated
exposure to painful stimuli may lead to the emergence of more
complex escape behaviors that are adapted to the specific

situation. The learning process may continue and enable responses
to nonpainful predictors of pain: fear replaces the perception of
pain as the driving force, and avoidance supersedes escape as the
response. Pain may reveal yet another dimension when substantial
tissue damage cannot be prevented: it produces the depressant
attribute of suffering, leading to inactivity that promotes
recovery from tissue damage. The current study is interested in
early conscious manifestations of pain that generate escape
behavior, but not in fear or recovery-related mechanisms.
Insight into the functional role of a system component can
be garnered by experimentally activating, inhibiting or even
destroying it, and measuring changes in overall system
performance. This approach of specifically targeted interventions
under otherwise controlled conditions is rarely possible in human
subjects or patients. Such experimental groundwork for future
therapeutic breakthroughs must be done in animal models. This
discussion will be limited to the rodent model, which is
economical, leads to behavioral test setups of manageable size
and benefits from a large existing knowledge base.
Pain is a private experience which cannot be measured
directly, either in humans, or in animals. It must be inferred from
behaviors. The behavioral vocabulary differs considerably
between species, from complex verbal behavior in humans to
simple vocalizations, escape reactions or autonomic responses in
animals. Pain tests for animals have been reviewed exhaustively
in the literature.1 -4 They can be classified according to two main

criteria: (1) type of stimulus applied, (2) type of response
Dubner2 has reviewed advantages and disadvantages of
different types of nociceptive stimuli, and thermal stimulation
emerged as a good choice. Thermal pain threshold and tolerance
levels are similar between species, promising validity of results
across species barriers. Some of the most popular rodent pain
tests use a thermal stimulus: the tailflick test,5 the hindlimb
withdrawal test6 and the hotplate test.7 The former two tests
commonly rely on a radiant heat stimulus. The absence of a
concurrent mechanical stimulus is thought to be an advantage of
radiant heat over other methods of thermal stimulation. However,
the temperature produced by this stimulus cannot be controlled,
monitored and maintained. The paw withdrawal and hotplate tests
keep the animals unrestrained and in control over the stimulus
duration. This minimizes stress, shortens the test-adaptation
time for the subjects and is ethically sound. However, the reflex-
based responses on these tests are not appropriate indicators of
pain sensations that may be modulated independently from spinal
or supraspinal reflexes.2,8-10 Paw licking, a response commonly
measured in the hot plate test,7,11 is a supraspinal,1 but not
necessarily a conscious, voluntary behavior. Paw licking is often
a component of grooming, which regularly occurs in chronic
decerebrate rats.12
Many popular methods for measuring the conscious aspects
of pain in animals have used an electrical aversive stimulus. The
advantage of electrical stimulation is that it can be turned on and

off instantly, making it easy to design automated testing setups.
This type of stimulation has met some reservations because skin
receptors are bypassed, and unnaturally synchronous afferent
firing patterns are generated.2 However, electrical stimulation is
capable of eliciting "natural" sensations of predictable quality,
when proper attention is given to the electrode tissue
coupling.4,13 Vierck and coworkers14 have developed a novel
operant escape paradigm for rodents in which electrodes are
individually attached to the hindpaws of the subjects and are
designed to control the path and density of the current within the
tissue. The animal escapes the nociceptive stimulus by pressing a
lever with a forepaw. This method allows laterality-specific
testing, but the restrained subjects need extensive adaptation
and training. Methods with more modest requirements for
adaptation and training commonly place the unrestrained animal
in a shuttle-box.15-17 The box features a metal grid floor which
is connected to an electrical stimulator. A shock scrambler18
changes the polarity assignments of the grid in a random fashion,
in order to prevent the subject from learning which bar
combinations are safe. The quality of the nociceptive experience
generated in a foot-shock shuttle-box is unpredictable, because
the characteristics of electrode-tissue coupling are not
controlled. The animal escapes or avoids the footshock by
crossing into the opposite compartment of the shuttle-box. This
gross motor response is easier to learn than bar-pressing,19,20
further reducing the time needed to train the animals. Some grid-
shock shuttle-box paradigms have added a light stimulus in the

safe compartment, allowing titration of the drive to avoid shock
against the aversive effect of bright light.21,22 Rodents exhibit
an innate aversion against open or brightly lit areas, which
reduces their exploratory behavior.23 In the absence of a strong
nociceptive stimulus, rats in a shuttle-box prefer the dark
compartment over its brightly lit counterpart. It seems logical to
conclude that bright light could be used to discourage avoidance,
if the objective is to measure escape.

Design Criteria for New Paradigm

The goal of this project was to develop a method for
measuring pain at the conscious level, using the rat model. The
design of the new paradigm incorporated the following
(1) Measuring conscious aspects of nociception necessitates
an operant response that is representative of direct sensory
experience and not of secondary motivators, such as fear.
(2) The paradigm must include controls for non-nociceptive
factors influencing behavior.
(3) The testing method must be respectful of the subject.
Stress and pain are minimized when the animal is given control
over the stimulus duration.

(4) The model must permit efficient use of resources:
minimal need for adaptation and training, the possibility of a
repeated measures design and a potential for automation.
(5) Stimulus-subject coupling must be simple and produce a
predictable and natural sensation, even under sub-optimal
conditions. This requirement can be met with a thermal stimulus.

Description of New Paradigm

The newly developed assay for conscious nociception
consists of two components:
(1) A hotplate-based test, subsequently referred to as the
Escapetest, designed to quantify operant escape and to provide
additional information about jumping/hindpaw-licking responses
that can precede escape.
(2) A shuttle-box test, called the Darkboxtest, which uses bright
light to elicit escape behavior. It serves as control for non-
nociceptive factors that could potentially influence performance
on the Escapetest.

Equipment for Escapetest

The setup (Fig. 1) consists of a rectangular transparent
Plexiglas testing chamber (inside dimensions 29 cm wide, 15 cm
deep, 26 cm high) on a commercial electrical hot plate (IITC
model 35). Automotive window tinting foil makes the inside of
the box reflective. The chamber features two compartments,
separated by an aluminum wall which is suspended from the top
cover with a hinge and ends 9.5 cm above the hotplate. The non-
reflective black hotplate surface makes up the floor of one of the
compartments. The other compartment features a small black
anodized aluminum platform (7.5 x 15 cm) protruding from one of
the sidewalls. This platform is normally oriented parallel to the
hot plate surface, 2 cm above it. It is not heated or temperature
controlled, but--due to the convective heat flow from the hot
plate below--is always slightly warmer than room temperature,
but never more than 30"C. A 50 W halogen light source is mounted
above this escape platform. A metal filter screen below the lamp
minimizes the amount of heat radiation. The brightness is 25 foot
lamberts (measuring light reflected from a standard white
surface) on the escape platform and 0.05 ft lamberts on the
hotplate surface The animal can escape from the hotplate by
climbing onto the platform in the brightly lit compartment. A
trial is initiated by briefly rotating the platform into a vertical
position, and--if the rat had chosen to occupy it--effectively
ejecting the animal back into the dark compartment and onto the
hotplate. The lightweight hinged dividing wall swings out of the


Figure 1. Escapetest apparatus

way when contacted by the rat. The swift translocation of the
subject onto the hotplate, results in instant onset of the heat
stimulus and clearly defines the beginning of a trial. This method
eliminates the need for handling the rat between each trial.
Handling can interfere with shuttle-box performance, because it
potentially disrupts spatial orientation.24

Test Protocol for Escapetest

Each trial has a duration of 30 seconds and begins and ends
with a cycling of the escape platform. The animal isn't removed
from the apparatus or handled by the investigator until 10 trials
are completed. Each series of 10 trials is preceded by a pretrial
period during which no data are collected. That is, the rat is
placed onto the hotplate and after a 10 second delay the
escape platform is made accessible; ten seconds later, the first
trial is initiated by cycling the platform. The pre-trial period
allows the animal to orient itself in the apparatus and assures
that the first trial, like all the others, is initiated by platform

Response Variables for Escapetest

Unlearned nociceptive behaviors are observed only when
they take place before--or instead of--the operant escape: the

delay between the beginning of a trial and the onset of the first
hindpaw-licking or jumping episode is recorded. The number of
trials of each 10-trial series with an unlearned behavior
preceding or replacing escape is counted.
The latency of the first operant escape onto the platform is
measured on each trial, and the frequency of escape responses is

Equipment for Darkboxtest

The apparatus (Fig. 2) consists of two chambers (each
measuring 22 x 22 x 24 cm). They are connected by a small
passageway which is accessible over an elevated platform.
Fluorescent lamps (GE 2D, 21W, 2700K, 1350 Lumens) are
mounted inside two of the walls and below the opaque floor of
each compartment. The outside walls and the top are made of
black Plexiglas. A temperature increase in the test chambers is
prevented by forced cooling of the lamps and electronic ballasts.
A three-position switch allows illumination of either the left
chamber, neither chamber, or the right chamber. The brightness is
20 ft lamberts in the lighted compartment and 0.2 ft lamberts on
the dark side. The brightness inside the apparatus is less than
0.005 ft lamberts when the lights are off in both chambers. Three
indicator lights, activated by weight-sensitive transducers,
inform the investigator as to where the rat is located (left
chamber, passageway or right chamber).



Figure 2. Darkboxtest apparatus.

Test Protocol for Darkboxtest

The protocol begins by placing a subject into the apparatus,
with the lights turned off in both compartments. No data are
collected during a pretrial period of 10s. Then, the first test trial
begins when the light is turned on in the chamber occupied by the
rat. This type of trial, which is characterized by the presence of a
light stimulus, will subsequently be referred to as a T-trial
After 70 seconds the light is turned off. This marks the end of the
T-trial and the beginning of a 70 second trial period where all
lights are out (to be called B-trial). T- and B-trials alternate
until 10 of each are completed. At that time the subject is
manually removed from the apparatus.

Response Variables for Darkboxtest

The latency of the first crossing through the passageway
into the opposite compartment, as well as the total number of
crossings during the 70 second trial period, are measured during
T-trails and B-trials.

The Darkboxtest as a Control Test

An inference of nociception from behavioral measures is not
possible without taking into account potential motor

impairments. The task for measuring motor performance must be
similar to the task for measuring nociception to be valid as a
control. The Escapetest as well as the Darkboxtest involve
stepping onto an elevated platform to retreat from an aversive
stimulus. The escape latency in the Escapetest is influenced by
two antagonistic driving forces. The motivation to escape from
the heat stimulus is opposed by an aversion to bright light.
Inferences about changes in thermal nociception depend on the
assumption that the other side of the balance, the aversion to
light, has not changed. The Darkboxtest--during T-trials--is used
to monitor the driving force of light on escape behavior. Thus, if
an experimental manipulation produces no effect on escape
latencies on T-trials but alters latencies on the Escapetest, then
an effect on pain sensitivity can be presumed independent of a
motor defect or a generalized effect on aversive motivation other
than pain.
The Darkboxtest--during B-trials--measures how often the
subject crosses between compartments in the absence of a light
intensity gradient. Locomotion under such conditions has been
proposed as a model for general motor activity level,23 which can
influence escape latencies by increasing the probability of all
motor responses.
In spite of similarities between the Darkboxtest and the
Escapetest setups, differences do exist. This is beneficial, when
it reduces the risk of interactions between the tests. When the
same subjects are trained on both tasks, they must be able to
distinguish between them. However, it can be a drawback if the

difficulty level of the two tests is not the same. The Escapetest
is a one-way shuttle test, because the response is always
directed in the same direction toward the same compartment. The
Darkboxtest, on the other hand, is a two-way test, because the
aversive light stimulus can occur in either of the two chambers.
A two-way shuttle-box does not permit an association between
"comfort" and a specific compartment or location. Some
experimental factors are known to have differential effects on
one-way and two-way tasks.24-26

Data Collection Method

The data collection is currently done manually, assisted by
a computer. A series of computer programs, written in Microsoft
Excel 4.0 macro language, alert the investigator when trials begin
and end, and--on key stroke--log the time of behavioral events
into a spreadsheet. Additional macro programs calculate
latencies, compute summary statistics and produce charts from
these data. The temporal resolution of the latency measurements
depends on the speed of the computer, and with our equipment
(Macintosh PowerBook Duo 270 and Power Macintosh 7100/66) is
slightly better than one second. The variability of the trial
duration has a range of zero to plus 2 seconds, if the timing
macros are run under MacOS version 7.5 or later.
The Escapetest method, in its current state of development,
has the disadvantage that it allows the subject to step back onto

the hotplate after escaping. The additional heat exposure can
potentially sensitize the animal for subsequent trials.
Furthermore, the animal is free to remain on the hot surface over
several trial periods without escaping. How should an escape be
interpreted when it is preceded by a trial period with no escape?
Should the latency be measured from the most recent platform
cycling (method 1), or from the platform cycling that followed
the last escape (method 2)? In other words, should cycling of the
platform mark the beginning of a trial only when the rat occupied
it, or be independent of the behavioral situation? Extreme
situations can be imagined where the answer to this question
would be relevant, but so far they have not occurred in any of our
experiments, as determined by separately evaluating and
statistically analyzing escape behavior by both methods. The data
presented in this report are based upon analysis method 1;
however significance is reported only when confirmed by analysis
method 2.
A technical solution to prevent the rat from touching the
hotplate after escape, or between trials, has been found and will
be implemented, along with computer-controlled automation of
the data collection.

Statistical Analysis of Individual Animal Data

Analysis of escape latencies. The influence of experimental
factors on the likelihood of escape (in Escapetests or

Darkboxtests) was assessed in individual animals via Cox
Proportional Hazards Modeling of the escape latencies. The Cox
model assumes the existence of a baseline or reference hazard
function (i.e., a profile of the probability of escape as a function
of time during the course of an individual trial under baseline or
reference experimental conditions). Underlying hazard functions
for various experimental conditions are reflected by the pattern
of escape or initial crossing latencies observed under those
conditions. Hazard function time intervals ranged from 0-30
seconds for thermal escape latencies in the Escapetest and from
0-70 seconds for light-induced escape latencies during T-trials
in the Darkboxtest. The Cox model further assumes that the
influence of experimental test conditions on the likelihood of
escape or initial crossing can be expressed as a proportional
increase or decrease in the baseline hazard function over its
entire range.
A Cox model computer algorithm estimates linear function
coefficients for indicator variables representing the presence or
absence of various experimental conditions so that they best
reflect the proportional change in the baseline hazard function as
evidenced by the difference in escape or initial crossing latency
patterns between baseline and experimental conditions. The
estimated coefficients can ultimately be re-expressed via
exponentiation as hazard ratios (i.e., the proportional increase or
decrease in the baseline hazard function attributable to various
experimental conditions).

A useful feature of the Cox model is its ability to
incorporate right-censored latencies into the hazard ratio
estimation process. Such right-censoring occurred when
individual animals failed to escape by the end of a 30-second
trial in the Escapetest or failed to cross by the end of a 70
second T-trial in the Darkboxtest. The statistical significance of
estimated hazard ratios was assessed by testing the
corresponding coefficient z-ratio for significant difference from
0 (the z-ratio is the estimated coefficient divided by its standard
error, assumed to follow a normal distribution with mean 0 under
the null hypothesis that a given experimental condition does not
significantly influence the baseline hazard rate).
Under certain circumstances, the hazard ratio could not be
estimated for a particular experimental condition due to right-
censoring of all latencies observed under that condition (this
occurred only in the 34C experimental condition for escape
latencies, although not universally in all animals). In these
instances, the score test was used to assess whether the
influence of this experimental condition on the hazard rate was
statistically significant.

Analysis of crossing frequencies in the Darkboxtest. The
influence of experimental factors on the frequency of Darkbox B-
trial crossing frequencies was assessed in individual animals by
comparing mean per-trial crossing frequencies among various
experimental conditions via independent groups analysis of
variance (ANOVA). For an individual animal, crossing frequencies

for individual trials were assumed to be independent
observations, grouped or completely cross-classified by the
levels of relevant experimental conditions. Main effect and
interaction F-tests were used to assess the overall significance
of effects attributable to various experimental conditions, while
contrasts of specific group means, also via F-test, were used to
assess the significance of differences between specific
experimental conditions of interest.
Due to a pattern of observed variability in which standard
deviations for B-trial crossing frequencies increased as
corresponding means increased, a second analysis was performed
in which crossing frequencies were transformed logarithmically
prior to analysis (since individual trial crossing frequencies of 0
were observed in all animals, 1 was added to each crossing
frequency before log-transformation). Resulting homogeneity of
group standard deviations and improvement in residual diagnostic
plots indicated the appropriateness of this transformation in
assessing the significance of effects due to various experimental
For ease of interpretation, the data (along with P-values)
are presented and charted in non-transformed form. However,
significance is reported only when confirmed by the analysis of
the log-transformed data.

Validation Experiments

Animal Subiects

One female albino Sprague Dawley rat (Rat #27) weighing
284 g was used for a preliminary Escapetest experiment. All
subsequent subjects were female outbred Long Evans rats
("hooded rats") obtained from Harlan Sprague Dawley. A group of
eight animals (group DR 04 consisting of rats #401-408) was
used to illustrate the relationship between temperature and
response latency. The data from six animals (group DR 05,
consisting of rats #502, 503 and 505-508) was used to
investigate the relationship between subcutaneously
administered morphine doses and escape latency. The morphine
sensitivity of the Darkboxtest was demonstrated in rats #44 and
45 (group PR 22). All the animals were individually housed in
wire mesh cages, under a 12h/12h light/dark cycle, with rat
chow and water available at libitum. Testing was conducted
during the second half of the light cycle, always by the same
investigator. The weight of the animals was between 255 and 299
g and was monitored regularly as an indicator of their well-being.
A steady weight gain was observed in all the subjects. The
experimental protocol was approved by the Institutional Animal
Care and Use Committee of the University of Florida.

Training of the Subjects

Training for the thermal escape task. Most subjects
required a total of about 18 training sessions (10 trials/session,
one session/day) to exhibit stable performance on the escape
task. The first five sessions were conducted with the light over
the escape platform turned off, to speed discovery of the
platform as a retreat from the stimulus. The temperature was
gradually increased from a non-nociceptive 34"C to a maximum of
50C during this initial phase of training. Subsequently, the light
was turned on, and the next phase began at 34C and progressed to
50C. Occasionally, a session was conducted at 34*C.

Training for the Darkboxtest. Training for the Darkboxtest
began after the subjects successfully acquired the thermal task.
The light-induced escape latencies reached an asymptotic
minimum after about 7 daily sessions (10 T-trials and 10 B-
trials each).

Validation of the Escapetest

The latency of operant escape represents the primary
response variable in the Escapetest. It is expected to change as a
function of stimulus intensity (temperature of the hotplate) and
to be sensitive to pain-reducing interventions, such as
administration of morphine. Retreat to the platform in the

absence of a nociceptive stimulus should be an infrequent event,
because it indicates that the response variable is no longer a
measure of escape alone, but of avoidance as well.

Temperature-latency relationship. Preliminary data were
collected from rat #27, with an earlier version of the Escapetest
that did not use a light source over the escape platform. It was
found that escape latencies appreciably changed within the
temperature range of 45" 50C but asymptotically approached
plateaus beyond these limits. Therefore, tests conducted with the
current design of the apparatus were limited to three
temperatures: 34", 48" and 50"C. Most subjects (group DR 04)
failed to escape during the 30-second trial periods at 34"C, but
did so regularly at 48"C and 50C. In the absence of an escape
response, a latency value of 30 seconds was assigned. Censoring
the response variable at this cut-off value limited variability at
the high end of the latency scale and led to a non-Gaussian
distribution of latencies (Fig. 3). At high temperatures, on the
other hand, variability was limited at the low end of the scale,
because the latency cannot be < zero. The result was a rightward
skewed distribution. This situation made the mean and standard
deviation, which are conditional on a normal distribution,
inappropriate choices for data reduction.27 A scatter plot of all
observed latency values was therefore used to illustrate
variability, and the median was chosen as the measure of central

Figure 3. Thermal escape latency distributions at different
temperatures. Columns: frequency; line: cumulated frequency;
20 trials/temperature.

RAT # 401












e- in (sec)

RAT # 408



-7%l ~7









.( I)
r .,o

/ ""t
f| |

* fO i*lll,,ii .,i, ii,, ,, W
0 S S S S


Figure 4 illustrates--in eight subjects--that the
Escapetest revealed a stimulus intensity-response relationship.
This meets an important criterion for behavioral tests of
nociception.28 All the subjects escaped during the majority of
the trials at 48 or 50C, indicating that the sensory experience at
these temperatures was painful.4,29

Sensitivity to morphine anti-nociception. The animals of
group DR 05 were subjects for an experiment using morphine
(1.25 mg/kg, 2.50 mg/kg) to demonstrate the sensitivity of the
Escapetest to anti-nociceptive interventions. Volume-matched
saline controls (0.20-0.22 ml) preceded and followed two
morphine doses, which were separated by a five day delay to
minimize the development of tolerance. A first series of ten
escape trials began 15 minutes after the subcutaneous injection,
a second series followed 12 minutes later. In two of the subjects
the effect of morphine was evident during the first series; in the
remaining animals the drug action peaked during the second
observation period. Morphine had an effect on the distribution of
escape latencies that was similar to a drop in test temperature
(see above): the variability became larger under the influence of
the opioid (Fig. 5). The median latency--after administration of
morphine--was consistently increased over the control value in
most animals in a dose-dependent manner (Fig. 6).
These experiments demonstrate that the thermal Escapetest
is able to detect anti-nociceptive effects of morphine doses

Q) O




0 V







o 4
u -c


0. Cu
in -^

" IU


^ g-S

(-S) I1 3S

I I II 1111 I
1 11111 1

(3S) 1B 3


( IS) IB ISa


(3i~) 0111

I Ir -n

(Os) 01 O

*,, ,s-

(oas) 01 'Sd

I III ,,


(3es) ol O











jouAnbaj m w


o .0

s I


,bwanbojy wm


:SZ |

. . fl

Auanbaj uno

: z 2



o mwnbo j

SZ 2

: oz I



.P I 0



e- t









( DN
nu -

I If C






0 a 0 0

,,,,i^ ,, ^ i i .^ -
I I 0
II ~ aI II t I II 111 0E
3 0

(r,) WI 0s ( s) ii .e ,


o S o o T o ....|....S|
i~E K -I ,.. j

I II II ii a .- "5

0E 0

(aas) e13 (:as) :le nosa n


"11 iII -I 111111
i Q
II I I I 111
0: 0: .-
0~ -
0 0 0 -0
C a a~
0 II 0 ,

11 II I

which have been shown to attenuate escape responses from other
forms of nociceptive stimulation.30

Avoidance detection in the Escapetest. Rats undergoing
Escapetests over weeks may learn to be afraid of the chamber
containing the hotplate as a potentially unsafe place. The operant
behavior might no longer be in response to a nociceptive
experience but to fear, and if so the assay ceases to measure
conscious nociception. Occasional tests at a neutral temperature
(e.g., 340C), when unpredictably scheduled, will reveal whether
avoidance learning took place. Platform utilization is likely to be
escape when it occurs regularly at high temperatures but only
infrequently in the neutral range. The introduction of a light
source over the escape platform was so effective in suppressing
avoidance, that only few examples are available to illustrate the
phenomenon. Rat # 304, when subjected to a first test at 34*C,
after learning to escape at higher temperatures, never left the
hotplate during the ten trials (Fig. 7). However, two weeks later,
after daily tests at nociceptive temperatures, it retreated to the
platform during half of the trial periods at 34C. This episode is
one of the few available for illustrating that occasional
Escapetests at 34C are useful to detect avoidance behavior.

Figure 7. Rat #304. Avoidance behavior at non-nociceptive
temperature of 34"C. No avoidance during test (1), but responses
were observed during test (2). Left: latency histograms. Right:
scatter plot with median latency. Ten trials/test.

34"C (1)

.. .. .. ... .. 0%

sc lat (sec)

34'C (2)






mmm_ 0|

sc tat (sec)

median median
30.00 sec 27.50 sec







Test 34C (1) 34C (2)



W 8

Validation of the Darkboxtest

Light-induced escape latency. The effect of morphine on
light-induced escape latencies (T-trials) was studied in subject
group PR 22. The latencies measured under control conditions (no
injection at all, or saline injection) and following morphine doses
of up to 5 mg/kg were similar (Fig. 8). However, an increase in
median and mean escape latency was observed when the morphine
dose was 10 mg/kg or more. High doses of morphine (2 10 mg/kg)
led to a pronounced increase in the variability of latencies (Fig.
Under baseline conditions (no drug), the median latencies in
the Darkboxtest were similar to those in the Escapetest at 50C
(compare Fig. 4 on page 29 and Fig. 8 on page 39). Escape behavior
in the Darkboxtest was much less sensitive to morphine than
latencies in the Escapetest. However, once the effective dose was
reached, morphine affected variability in a similar manner on
both tests (compare Fig. 5 on page 31 and Fig. 9 on page 41).

Number of crossings between dark compartments. The
average number of crossings during 70 second B-trials was small,
leaving little room for a depressant effect by morphine (Fig. 10).

Usefulness as a control test. The Darkboxtest is useful in
combination with the Escapetest, because it reveals whether
factors unrelated to nociception modulate escape behavior. The

Io o
4- N

a )

r- U i C

o .g .
.o-0 o Eo

-- a >Se

.o -.

'0 ,-- -

E- a --
1 -0 (

II -0 -
-o a-Q e

r- E

. cc 0 5
Ch 0

.- U E-

mLi S1 a
11-~rO ~C
V~C 0
mcQ 0.0
0)2a! E
irDS U


Ill les

I III I ou

I II I ou
I I les


SI l- O'S

-I I I II leS


11 les

II II I I ou

I II ou

0 0 0 0 0 0 0
(Ms) lt| 3Sa

I I I I ou

II les



I I II ou


(3as) lel :,s

Figure 9. Distributions of light-induced escape latencies after SC
morphine injection. Ten trials/dose. There was a delay of at least 6
days between morphine tests, in order to minimize the development
of tolerance.

RAT # 44

morphine 2.5 mg/kg




sC at (Ec)

morphine 5.0 mg/kg

morphine 10 mg/kg











69 0%

RAT # 45

morphine 2.5 mg/kg
10 1009%

8 80%

6 60%

4 40%

o2 .0%

esc tt (ec)

morphine 5.0 mg/kg

emc 63t (9ec)

morphine 10 mg/kg

8 0.1

14 0.
2. 0

o o oo0



m.E .
4-,o -

0 0c

C 0
U0 b

ao : -
C 0
0 C

a0 p
C ^

0 am4

E cE
S o v0i

Z E cE

-o m

E _c

I Is


I I ou \

I sl

I |es

I ou

pouad eul 30s OL / sumssoj o jo jqujnu

I leOZ

I I es



I S '2s


poualeeul S OZ / s6ulssoJ, o oqumnu

similarity in response characteristics (e.g., median latency and
latency distribution for nearly identical responses to different
aversive stimuli) between the two tests strengthens confidence
in the Darkboxtest as a valid control test. In the case of
morphine, light-induced escape was affected only by large doses,
permitting the conclusion that the effect of lower doses on
thermal escape resulted from attenuation of thermal nociception
and not from reductions in light-induced aversion or motor


The Escapetest/Darkboxtest paradigm is a promising new
addition to the inventory of pain measurement tools. It uses a
thermal stimulus and an operant escape response to measure
conscious aspects of nociception. It is sensitive to the intensity
of thermal stimulation and to the anti-nociceptive action of
morphine. It provides controls for extraneous factors not related
to nociception, such as avoidance learning, motor impairment, or
a generalized effect on aversively motivated behaviors. The tests
are ethically sound, because the subjects are unrestrained and
can terminate the aversive stimulus by escaping.



Information regarding the functional role of a brain site can
be garnered by experimentally destroying it and measuring
resulting behavioral changes. The interpretation of results is
complicated when lesions are off target, or not limited to the
intended structure. Such errors may be caused by inaccurate
methods of approaching the intended structure, or poor control
over the extent of the lesion. The goal of the present work was to
identify factors that contribute to inaccurate placement of
lesions and to find methodological refinements that minimize
these errors. Specifically, the aim was to maximize the yield of
successful thalamic lesions in rodents by eliminating as many
sources of error as practical.

Stereotaxic Methodology

Many factors that can compromise the accuracy and
reproducibility of experimental brain lesions were identified as
early as 1908 by Horsley and Clarke.31 These authors realized
that precise navigation to any given point within the brain was
not possible with manual guidance, aided by a few skeletal
landmarks, and they were the first to define cerebral space by
means of a three-dimensional rectangular coordinate system.
They further designed mechanical hardware for holding the
cranium of the subject in a defined position and mechanically
advancing lesion needles to the target. The combination of a
coordinate system and associated hardware is now referred to as
a stereotaxic system. A stereotaxic procedure begins with
determination of the target coordinates in a reference specimen
or atlas. Linear drives, calibrated in units of distance will then
direct the lesion probe to the chosen location.

Coordinate System

Coordinate systems can be Cartesian (rectangular) or polar
(angular). Stereotaxic methodology frequently uses a combination
of both, in order to provide alternate approach paths to a given
target (e.g., for avoiding large blood vessels).32 Cartesian
coordinate systems describe the position of a point of interest as

distances from three perpendicular reference planes. Readily
identifiable landmarks with minimal inter-subject variability are
needed to define these planes. External skeletal,31,33-35 as well
as intracerebral radiological landmarks36,37 have been used for
this purpose, and the choice is often dictated by species-specific
anatomical circumstances. The present work is focused on a
rectangular coordinate system adapted for the rat by Paxinos and
The coordinate system for the atlas by Paxinos and
Watson35 relies on the cranial landmarks Lambda and Bregma, and
the centers of the external auditory meati for defining horizontal
and frontal reference planes. The midsagittal plane represents
zero in the lateral dimension of the coordinate system. The
authors of the atlas chose a horizontal orientation of the Lambda-
Bregma axis (flat-skull position) because craniometric
comparisons have shown that this alignment minimizes
stereotaxic errors due to strain and weight differences between
subjects and the atlas reference.38,39 The atlas offers
alternatives of either the horizontal plane through Bregma or the
interaural horizontal plane as the origin for vertical coordinates.
Likewise, two reference planes are used for the antero-posterior
dimension: the frontal plane through Bregma and the frontal plane
through the interaural axis (Fig. 11).

Figure 11. The atlas of Paxinos and Watson defines the stereotaxic
coordinate system for rats as follows: the head is held in the
stereotaxic frame by earbars so the interaural axis (through the
centers of the left and right external auditory meatus) is horizontal.
Lambda and Bregma are skeletal landmarks: the head must be rotated
around the interaural axis until the Lambda-Bregma axis is horizontal
(flat-skull position). Coordinates of brain sites are measured as
rectilinear distances from perpendicular reference planes.
Two horizontal reference planes are available for defining vertical
coordinates: the horizontal plane through Lambda and Bregma and the
horizontal plane through the interaural axis. Likewise, two frontal
planes can serve as zero reference for antero-posterior coordinates:
the frontal plane through the interaural axis and the frontal plane
through Bregma. Modified (with permission) from Paxinos, G. and
Watson C. The rat brain in stereotaxic coordinates. 2nd ed. San Diego:
Academic Press, 1986

Interaural Bregma
frontal plane frontal plane

Stereotaxic Frame

Lesion probes are moved to the desired coordinates within
the brain by means of linear microdrives. The stereotaxic frame
is the mechanical device that aligns the degrees of freedom of
the microdrives with the axes of the coordinate system. Horsley
and Clarke31 were the first to build a stereotaxic machine.
Numerous adaptations followed to accommodate a variety of
species, including human patients.32,33,36,37
Stereotaxic equipment for rodent surgery is available from
a number of manufacturers. The products differ in technical
details, but they are all compatible with the coordinate system
used by Paxinos and Watson.35 The Kopf Model 963 Stereotaxic
Instrument was used in the present rodent study.

Stereotaxic Errors

The accuracy of lesions may be compromised by the
following factors:32,40

(1) Bias due to anatomical variability of the landmarks used
to define the coordinate system.
(2) Bias due to misalignment between the stereotaxic frame
and the coordinate system.

(3) Equipment-related bias--e.g., due to imperfections in
the attachments of electrode holders that become manifest when
electrodes are changed.
(4) Errors due to distortion of neural tissue.

Bias due to anatomical variability. Assembly of
anatomically homogeneous experimental groups in order to
minimize variability is more feasible in the rodent than in other
species. However, there may be reasons for using rats that differ
in sex, strain or weight from the atlas model. The former two
variables have only a minimal effect on stereotaxic accuracy, as
long as a coordinate system based upon the flat-skull position is
used.35,38,39 Differences in weight, on the other hand, may
deteriorate accuracy in the antero-posterior dimension
considerably, if no corrective measures are taken. Whishaw and
coworkers39 found that a linear relationship exists between the
weight of the animals and the length of the cranium. The use of
atlas-derived antero-posterior coordinates without compensation
for the larger skull size of older rats leads to biased targeting.
Only a few studies of craniometric variability exist for
rats, and they focus almost exclusively on the relationship
between head size and stereotaxic targeting accuracy in the
antero-posterior dimension, where the largest errors occur.
Stereotaxic accuracy is limited by the reliability of skeletal
landmarks, which are used to predict the location of brain
structures. Bregma is a landmark which is located anteriorly on
the skull, and it is often defined as the juncture of the

bregmoidal suture with the sagittal suture. In most rodents, the
left and right half of the bregmoidal suture meet the sagittal
suture in separate locations. It has been shown that the average
of these two positions is less variable than each individual
juncture.41 Lambda, the posterior landmark on the surface of the
skull, is defined as the intersection between the lambdoid and
sagittal sutures. Paxinos et al.35,38 discovered that curves of
best fit along the lambdoid or bregmoid sutures intersect the
midline at points which are more stable than the bony junctures,
and they adopted these reference points for their atlas (Fig. 12).
Slotnick and Brown41 found that of the three landmarks--
Bregma, the interaural midpoint and Lambda--only the first two
are reliable predictors for the antero-posterior dimension. For
best results, it has been recommended to use Bregma as the
reference for rostral stereotaxic targets and the interaural
reference to define the position of caudal structures.38 The atlas
of Paxinos and Watson35 facilitates this approach by providing
the coordinates from both referents. Further improvement of
antero-posterior stereotaxic accuracy is possible when atlas-
derived coordinates are adjusted for differences in head size
between the subject and the atlas standard. The distance between
the frontal plane through the interaural axis and the frontal plane
through Bregma, with the head aligned so that Bregma and Lambda
are in the same horizontal plane, is a measure commonly used for
head size comparisons (subsequently to be referred to as I-B). The
difference between I-B of the subject and I-B of the rat atlas has
high predictive value for the position of brain structures in

Figure 12. Landmarks for the stereotaxic coordinate system in
rodents. sL = Lambda, defined as the juncture between lambdoid and
sagittal sutures; L = Lambda, defined as the intersection of the
curve of best fit (dotted) along the lambdoid suture (gray) and the
midline (dotted); B = Bregma is defined as the intersection of the
curve of best fit (dotted) along the bregmoid suture (gray) and the
midline (dotted); M = the straight section of the sagittal suture used
to define the midline. Modified (with permission) from Paxinos, G.
and Watson C. The rat brain in stereotaxic coordinates. 2nd ed.
San Diego: Academic Press, 1986

rodents.41,42 This difference, when used as the variable in a
linear regression function, reliably predicted the location of
targets along the entire length of the brain in a craniometric
study of 14 albino rats by Slotnick and Brown.41 The authors used
the regression equation hs = ha + b(x-xn), where hs was the
predicted antero-posterior coordinate, ha the atlas-derived
coordinate, x the subject I-B, xn the representative craniometric
average of I-B and b the slope coefficient (a 0.6). The interaural
reference was used in this study. In strains where similar
craniometric data do not exist, a scaling factor can be used to
compensate for differences in head size, with little sacrifice in
predictive value for targets located between the Interaural and
Bregma reference.41 In the present work, I-B of the subject was
determined by surveying the relative locations of the two
landmarks with a stereotaxically guided needle probe. The factor
(Fa) was then calculated with the formula Fa = (Bh- Ah)/xa ,
where Bh is the reading on the antero-posterior microdrive scale
at Bregma, Ah the reading at the interaural point and xa the atlas
I-B (9.0 mm in the atlas of Paxinos and Watson35).
Weight-related variability of head size is less pronounced
in the vertical dimension. Resulting errors can be reduced with
the same methods described for the antero-posterior axis.
Improved results can be expected when coordinates are measured
from the closest reference plane, i.e., from the Lambda-Bregma
horizontal plane for dorsally located structures and from the
interaural horizontal plane for deep targets (Fig. 11, page 48). The
distance between the two referents--with the head aligned so

that Bregma and Lambda are on the same horizontal plane--will
be referred to as I-L. A scaling factor based on l-L may be
calculated with the formula Fv = (Lv- Av)/ya where Lv
represents the reading of the vertical microdrive at Lambda, Av
the reading at the interaural midpoint and ya the atlas derived I-
L. Some investigators39 prefer to measure the depth of the
electrode penetration from the surface of the dura, rather than
from the skeletal reference point because the thickness of the
cranial bone increases significantly as the rats age.
No craniometric studies could be found that dealt with the
anatomical variability in the lateral dimension. Lateral
coordinates are measured from the midsagittal plane, which can
be defined by three landmarks: the interaural midpoint, the
posterior linear portion of the sagittal suture (M in Fig. 12, page
52) and the center of the sagittal venous sinus. In the present
work the reference plane was defined by the average position of
all three landmarks. The average of the antero-posterior and
vertical scaling factors has been used to adjust lateral
coordinates for non-standard head size. A systematic study of the
consequences of this approach has not been conducted.

Misalignment of coordinate system. Anatomical variability
of landmarks and procedural errors can cause misalignment
between the coordinate system and the brain of the subject. The
focus of the present report is on alignment errors that can be
avoided through proper techniques.

The stereotaxic procedure begins with the placement of the
earbars. A rare but irreconcilable error occurs when the auditory
meati of the subject are asymmetrical, due to anatomical
variation.31 Improper positioning of the earbars is more common
and can readily be detected and remedied. A common mistake is to
place the earbars too far anteriorly, in the temporomandibular
joint, instead of the auditory meatus. This type of error must be
suspected when I-B or I-L differ more than approximately 20%
from the norm listed for adult rats of the chosen strain.35,38
Proper placement in the auditory meatus usually brings the
earbars into a position slightly posterior to Lambda (Fig. 13).
Once the head of the rat is held in the stereotaxic frame by
the earbars, it can be rotated in the sagittal plane around the
interaural axis. The coordinate system used in the atlas of
Paxinos and Watson35 requires that the head is held in a position
where the Lambda-Bregma axis is horizontal (flat-skull position).
This alignment can be achieved as follows: (1) Preliminary
adjustment of the incisor bar to a setting based upon
craniometric averages.38 (2) Survey of the relative vertical
position of Lambda and Bregma in the stereotaxic frame, using a
needle probe and the millimeter scale on the microdrive. (3) The
survey result is then used to calculate an angular correction for a
better approximation of the flat-skull position (Fig. 14). (4) This
angle, the preliminary setting of the incisorbar and the distance
of the bar from the interaural axis are the basis for computing a
new setting on the linear scale of the incisorbar support (Fig. 15).

correct incorrect
earbar position

Figure 13. Verification of the earbar position. (1) Adjust the
incisorbar to an average setting appropriate for the chosen rat
strain (-3.9 mm for Long Evans rats), in order to approximate the
horizontal orientation of the Lambda-Bregma axis. (2) Verify the
horizontal distance between the earbar and Bregma (I-B): a
measurement within w20% of the craniometric average for the
chosen strain suggests correct placement of the earbar. (3) Verify
the vertical distance between the earbar and Lambda (I-L): a
measurement within 20% of the craniometric average for the
chosen strain suggests correct placement of the earbar. Common
error: placement of an earbar in the temporo-mandibular joint (TMJ)
instead of the auditory meatus. Modified (with permission) from
Paxinos, G. and Watson C. The rat brain in stereotaxic coordinates.
2nd ed. San Diego: Academic Press, 1986

use of needle probe and microdrive for surveying
the vertical position of bony landmarks

0 0

Lambda H Bregma
(negligible change
of vertical position)

i( y' yn y

Interaural WN

gray: intended horizontal alignment (flat-skull position)
Xn, yn: craniometric average for I-B and I-L of chosen rat strain
r: distance of Bregma from Inter-aural axis of rotation
dashed: actual alignment
L,B: microdrive readings obtained during survey of landmarks

geometric relationships:

B L = dy (difference in vertical position of Bregma and Lambda)
y = yn+ dy
x = V(r2 y2)
W, = Arctan(xn, Yn); W = Arctan(x, y) ; [Arctangent of coordinates x, y]
dw = Wn -W
dw is the angular correction needed to achieve the flat-skull position

Figure 14. Survey of bony landmarks Bregma and Lambda to compute
the angular correction in head position leading to the "flat-skull

Failure to adjust the head of the subject to the flat-skull
position results in horizontal as well as vertical targeting errors
(Fig. 16). The horizontal error is related to the depth of vertical
penetration. The vertical error is a function of the antero-
posterior coordinate. It can be compensated for by adjusting the
vertical reference plane accordingly.

Equipment-related stereotaxic bias. The survey of skeletal
landmarks, as described above, is best performed with a sharp
needle, and not with the radiofrequency electrode, which has a
rounded tip and is too costly to be put at risk. This means that the
probe and its holder need to be removed from the apparatus and
replaced after the preparatory steps are completed. This not only
introduces a bias due to imperfections in the mechanical
attachments, but it also leads to a loss of the relationship
between reference planes and the microdrive scales. To overcome
this problem, an arbitrary skeletal landmark is identified on the
skull, and its coordinates are measured in all three dimensions
with the surveying needle. The microdrives are recalibrated to
this arbitrary reference point after the switch to the new
A radiofrequency lesion extends radially from the exposed
tip of the electrode and not axially. The largest radial lesion
extent is expected to be near the center of the exposed tip
length.43,44 The vertical stereotaxic coordinate must be
increased by one half of the exposed tip length to avoid a dorsal
stereotaxic bias.


Interaural V bar

I<- D > -

gray: intended horizontal alignment (flat-skull position)
dw: angular correction needed to achieve flat-skull position
in: preliminary setting of incisorbar (based upon craniometric averages)

dashed: actual alignment (nose down)

geometric relationships:

Vn = Arctan(D, in) [Arctangent of coordinates D, in]
V =Vn+dw
i is the new setting for the incisorbar that will establish the flat-skull position

Figure 15. Calculation of linear incisor bar setting for achieving
flat-skull position

Interaural _c TI-T

dashed: head not accurately aligned to flat-skull position (nose down)
dw= angular head alignment error (see Fig. 14)
x= Interaural-Bregma distance (I-B)
h= antero-posterior coordinate of target (relative to Bregma)
s= vertical coordinate of target
P= electrode penetration point
Ta= target as chosen in atlas
T= actually reached target
P-Ta (gray): intended electrode track
P-T (black): actual electrode track
antero-posterior targeting error at vertical coordinate s : e= s sin dw
vertical targeting error at antero-posterior coordinate h (Lambda used
as vertical reference): dyh= (h/x)-dy

Figure 16. Horizontal and vertical targeting errors resulting
from failure to adjust head to flat-skull position.

Errors due to distortion of neural tissue. The tips of
radiofrequency electrodes are dull and will not readily penetrate
the dura mater without compressing the brain. The meningae must
be cut prior to electrode insertion, and the electrode must be
advanced slowly to minimize stereotaxic bias due to distortion of
the tissue.
Swelling of the brain can have a deleterious effect on
stereotaxic accuracy, especially when the depth of electrode
penetration is measured from the surface of the dura mater. The
problem can be minimized by premedicating the animal--e.g., with

Computer Aided Methodology

A spreadsheet-based computer program has been designed
for obtaining the incisorbar setting needed for proper alignment
of the subject's skull, calculating scaling factors, adjusting the
vertical coordinate for electrode tip dimensions and translating
coordinates into microdrive settings. The software alerts the
investigator when the earbars appear to be inserted incorrectly--
e.g., in the temporomandibular joint instead of the auditory
meatus. The program provides boundary coordinates for drilling
through the skull and recalculates the microdrive settings after
the transition from the surveying probe to the lesion electrode.
The computer calculates anesthesia doses--based upon the

weight of the animal--and automatically maintains a
standardized, time-stamped surgical progress report.
The stereotaxic procedure begins by centering the earbars
in the stereotaxic frame, leaving a gap of 0.2 mm between the
tips. A needle probe is used to determine the stereotaxic position
of the interaural midpoint. Subsequently, the rat is mounted in
the frame and the incisorbar is adjusted to the craniometric
average setting for the chosen strain. The skeletal landmarks are
surveyed stereotaxically with the needle. The incisorbar is
adjusted to the setting calculated by the computer for achieving
the flat-skull position. The needle probe is used for picking up the
new positions of the skeletal landmarks and for defining the
arbitrary reference point. The skull is trepanned at locations
defined by the program. The surveying probe is exchanged for the
lesion electrode. The electrode tip is manipulated to the arbitrary
reference point, and the new coordinates are entered into the
computer. The computer then provides the microdrive settings for
all lesion loci. Keystrokes direct a macro program to enter lesion
coordinates, RF energy levels and time of the day into the
progress report.

Lesion Methodology

Correlating lesion sizes and locations with behavior
represents a method of functional brain mapping. The resolution
of the functional map depends on the anatomical selectivity of
the lesion medium. At the macroscopic level, the lesions should
be tightly focused around the target. At the cellular level, they
should be selective for either gray or white matter. No method is
currently available that provides high selectivity in all respects.
The destructive effect of excitotoxic lesions is biased towards
gray matter, but it is poorly focused at the macroscopic level.
Electrical lesions have been refined to the point that they are
highly focused macroscopically, but they remain indiscriminate
at the cellular level. The present report discusses methodology
that uses electrical energy to produce lesions of predictable size
and shape.
Sellier and Verger45 pioneered the use of direct current
(DC) to produce electrolytic brain lesions. However, they provided
little methodological detail. Horsley and Clarke31 conducted the
first in-depth investigation of the mechanisms of electrolytic
lesions. They concluded that tissue destruction is caused less by
resistive heating than by the products of electrochemical
reactions at the electrode, including gas bubbles which lead to
pressure damage. The authors observed that anodal lesions tend to
be more focused than cathodal lesions, because considerably less
gas is produced at the positive electrode. The electrolytic

methodology has the drawback that lesions are often irregular in
shape and size, due to the unpredictable spread of gas bubbles and
toxic ions.46 The size of electrolytic lesions may vary by a factor
of four for a given set of lesion parameters.47 Electrolysis of
tissue fluids and gas formation are avoided when alternating
current (AC) is used instead of DC. The damage caused by AC
results from heat, which is generated in the tissue by oscillating
ionic movement. Lesion generators typically operate in the radio-
frequency (RF) range, between 300 and 500 KHz, well outside the
range of neuronal excitability. RF methodology is superior to its
electrolytic counterpart, because it leads to more distinct and
predictable lesions. It was the method of choice in the present

Radiofrequency Lesions

Electrodes. A monopolar electrode arrangement is used for
most deep RF lesions. The tip of the focal electrode is
stereotaxically centered within the target structure, while the
dispersive electrode is placed outside the central nervous system
to complete the electrical circuit. The current between the focal
electrode and brain tissue is mainly resistive at frequencies
below 1 MHz and--due to the small size of the exposed tip--of
high density.48 The high current density leads to considerable
heat dissipation within the tissue near the electrode tip. The heat
production per volume unit of tissue drops as the 4th power of

the radial distance from a tip of spherical shape.49 This leads to
a very narrow transition zone between necrotic and healthy
tissue. The dispersive electrode must be at least 100x larger than
the focal electrode, in order to disperse the current density
outside the target (Fig. 17). Foci of high current density and
unintended heat damage in adjoining tissue may result when the
contact area of the dispersive electrode is too small. Insulated,
capacitively coupled dispersive electrodes eliminate this risk, at
the expense of coupling efficiency.50 In small animal research,
tissue retractors or alligator clips which broadly contact the
moistened borders of the scalp incision serve as effective
dispersive electrodes.

Factors affecting lesion size. Temperature, the net result of
heat production and heat loss, is the most critical lesion
parameter. The temperature gradient at the focal electrode is
steep and depends on the magnitude of the RF current, the
electrode tip size and shape, the duration of the current and the
heat conductivity of the tissue. The boundary between reversible
and permanent tissue damage lies between the 45" and 50C
isotherms.47,48,51,52 Neural function is only temporarily (for
minutes) suppressed between 42.5" and 44"C.47,48,53
Resistive heat production is a function of current density,
which in turn is related to electrode surface area and tissue
impedance. Tissue impedance depends on fluid content and ionic
mobility and thus varies between tissues of different
vascularity--e.g., between gray matter and white matter. White


prolate spheroid
shape of lesion


Figure 17. Principles of the radiofrequency lesion method.
High current density near the focal electrode produces heat
and tissue damage. The current density and heat generation
drops sharply with increasing distance from the focal
electrode. The lesion has the shape of a prolate spheroid
and does not extend beyond the electrode tip. The coupling of
the dispersive electrode may be galvanic or capacitive.

matter, due to a low vascularity, is slightly more sensitive to RF
lesioning than gray matter.43 Ionic mobility changes throughout
the lesioning process: it increases as the temperature climbs to
60C, resulting in an impedance drop, then rises sharply when
higher temperatures are reached, and coagulation closes ionic
Heat losses are most pronounced in the vicinity of blood
vessels or CSF spaces and have a distorting effect on the
isotherm pattern and thus the shape of the lesion. Thermal losses
influence the electrical energy requirements for reaching and
maintaining a certain temperature.47 This leads to the conclusion
that electrical lesion parameters (amperage, voltage) are poor
predictors of lesion size, and that direct temperature feedback is
essential for the best results. Makers of state-of-the-art RF
lesion equipment meet this requirement by incorporating
miniature temperature sensors into the electrode tips.
Pecson and colleagues51 maintained a 65*C tissue
temperature for multiples of 2.5 seconds, in order to study the
effect of time on lesion size. They found a non-linear relationship
between the two variables: the lesion size doubled between 2.5
and 30 seconds, but increased only minimally between 30 and 75
seconds. Cosman et al.47 similarly reported that near maximal
lesion size was achieved by maintaining 75C for 30 seconds.
The electrode dimensions have an effect on the shape and
size of the lesion.44,47,52 The lesion extends radially outward
from a cylindrical electrode tip, and axially onto the insulated
portion of the electrode, but not downward beyond the tip. In the

transverse dimension a lesion diameter of twice to four times
the tip diameter can be expected at 70-80C. The shape of the
coagulated area resembles a prolate spheroid (Fig. 17, page 66).
Overheating may have unpredictable consequences: it may
lead to large and irregular lesions, due to explosive steam
formation, or it may result in undersized lesions due to the self-
limiting process of charring.43
A lesion time of 60-90 seconds, one fourth of which is used
to gradually approach a final temperature of 60-80C, is common
for producing large lesions.


It is important to conduct research in a manner that
efficiently utilizes resources. It is a common misconception that
a stereotaxic atlas can only provide some initial guidance for an
iterative trial and error approach to a lesion target. The present
review of relevant literature and a discussion of geometric
principles that underlie stereotaxic methodology should have
demonstrated that many sources of stereotaxic errors can be
eliminated by attention to detail and a few simple calculations.
The availability of spreadsheet programs makes the necessary
computations easy. Even the most accurate stereotaxic
procedures will not lead to the desired lesion results when


outdated lesion methods are used. Radiofrequency lesioning is
superior to electrolytic techniques and therefore has become the
method of choice in the neurosurgical operating room.
Radiofrequency equipment is available for animal research, and it
is a worthwhile investment, because more precise and thus more
easily interpretable lesion results can be obtained with fewer
animal subjects.



Diencephalic nuclei are way stations for many
paucisynaptic and polysynaptic nociceptive channels of different
origins. These nuclei represent nodes in a complex network of
sensory, motor and limbic interactions for which the anatomy and
physiology are much better understood than the function. The
development of treatments for intractable pain conditions would
benefit from a better insight into the function of thalamic
nociceptive systems. Clinicians would especially welcome better
methods for selectively reducing the unpleasantness of pain,
while preserving other aspects of the sensation. The present
study focuses on the rodent parafascicular thalamic region, for
which a role in the processing of aversive attributes of pain is
suspected. The evidence consists of anatomical tracing results,
physiological characterization of the neurons involved and some
functional data. Behavioral studies, using not only nociceptive but
also nonpainful aversive stimuli, are needed to better define the

sensory and/or motor functions of this nucleus in situations
involving aversion.

Anatomical and Physiological Background

General Topographic Anatomy

The center median/parafascicular complex--which is
distinguishable from other intralaminar structures by a lack in
immunostaining for calcium-binding protein (CaBP)54--is located
in the posterior diencephalon. The parafascicular nucleus
represents the medial component and the center median the
lateral component of the intralaminar complex of nuclei. The
parafascicular nucleus is found in all mammals, including
rodents, while the center median nucleus is highly developed only
in primates. The evolutionary development of the center median
nucleus from a small mass of cells between the rostrolateral end
of the parafascicular and the paracentral nuclei to a highly
differentiated structure parallels that of the putamen.54,55
The medial neighbor of the rodent parafascicular nucleus is
the mediodorsal thalamic nucleus, except caudally, where the
periaqueductal gray occupies this position. The dorsal border is
formed by the lateral habenular nucleus and the central lateral
thalamic nucleus. The latter adjoins the parafascicular nucleus

rostrally as well. The posterior and ventrobasal thalamic nuclei
are laterally adjacent. The CaBP immunopositive
subparafascicular nucleus is located below the caudal part of the
parafascicular nucleus, above the medial lemniscus.
The medial part of the parafascicular nucleus is perforated
by the fasciculus retroflexus. This bidirectional, neurochemically
heterogeneous (e.g., cholinergic and dopaminergic) pathway, which
is also known as the habenulo-interpeduncular tract, connects the
habenular nuclei with the basal midbrain--the interpeduncular
nucleus, substantial nigra (pars compacta, dorsal raphe and
tegmental nuclei.56,57 The fasciculus retroflexus is a
caudoventral extension of the stria medullaris, which originates
in the septum, entopeduncular nucleus and lateral hypothalamus.
The fasciculus retroflexus has been implicated in reward
mechanisms,56,58 inhibition of tonic pain59 and in spatial
memory.60 Thus, involvement of the fasciculus retroflexus could
be a critical determinant of effects on the behavioral assays
commonly used in thalamic lesion studies.

Anatomy and Physiology of Peripheral Inputs

Direct and indirect afferents from the periphery. The
existence of direct spinal afferents to the center
median/parafascicular region has been advocated by some authors
and denied by others.61-63 Those that may exist are likely to
originate in upper cervical segments (for discussion see Giesler

and coworkers64). Input to the center median/parafascicular
complex does not depend on the major ascending spinal pathways
(i.e., the medial lemniscus or the lateral spinothalamic tract):
Shafron and Collins65 evoked activity in this nucleus by
peripheral nerve stimulation in cats when all the spinal cord but
the ipsilateral anterolateral column was severed. Bowsher et
al.66 noted that bilateral section of the anterolateral tracts
interrupted signal transmission to the center median nucleus of
Retrograde degeneration62 and labeling studies67 in
rodents and subhuman primates,68 physiological experiments in
cats66 and conduction velocity measurements in humans69
suggest that inputs from the periphery to the center
median/parafascicular complex are largely polysynaptic, through
the cervical spinal cord, the medullary and mesencephalic
reticular formation, the reticular thalamic nucleus and the zona
incerta. The zona incerta acts as a relay station for signals from
the mesencephalic parabrachial and peripeduncular regions.67 The
nucleus gigantocellularis--a component of the medullary
reticular formation--shares many neuronal response properties
with the center median/parafascicular thalamus61 and appears to
be a critical relay for nociceptive inputs enroute to the center
median nucleus. Cooling the gigantocellular nucleus abolishes
activity evoked in the center median nucleus by nociceptive
stimulation.66 The relationship between the two structures
appears to be bidirectional, as toothpulp-evoked activity in the
nucleus gigantocellularis of the cat can be modulated (both, in a

facilitatory and inhibitory manner) by electrical stimulation in
the center median nucleus.70

Neuronal response properties. The nociceptive neurons in
the parafascicular nuclei are generally characterized by large,
often bilateral receptive fields, an absence of somatotopic
organization, a lack of intensity encoding, high thresholds for
excitation, and the presence of a long afterdischarge.61,69,71-73
Pearl and Anderson74 found that 70% of cells isolated in
the center median nucleus of cats responded to electrical
toothpulp stimuli. The receptive fields were large, usually
bilateral, with a crude topographic organization: dorsally located
cells responded to maxillary stimulation, and ventrally located
cells responded to mandibular tooth pulp stimulation. Cells with a
spinal receptive field were more common in the fringe area of the
parafascicular nucleus, or beyond its borders. Perl and
Whitlock73 found that unitary activity in the feline and primate
center median nuclei was elicited only by nociceptive peripheral
stimulation. Yen et al.71 conducted extracellular recordings in
the rat and found neurons in the parafascicular nucleus that
exhibited excitatory responses to nociceptive radiant heat
stimulation of the tail.
Peschanski and colleagues75 reported the presence of high
threshold (49-50C), short latency (30-40 ms) thermally-
activated neurons in the caudal intralaminar region. Some units
could be characterized as detectors of change within the
nociceptive range, without encoding absolute intensity and

somatotopic information. This prompted the authors to propose a
non-specific alerting function. Bushnell and Duncan76 found that
activity in the simian center median/parafascicular nucleus,
evoked by peripheral thermal stimulation, did encode intensity
and was sensitive to attentional states, but not to the presence
or absence of a behavioral response to the stimulus.

Central Fiber Connections

Telencephalic and cerebellar inputs. Cornwall and
Phillipson,67 after conducting retrograde tracing experiments,
concluded that the parafascicular nucleus represents a node
where inputs from the limbic-, motor- and reticular activating
systems converge. They demonstrated direct afferents from the
orbital and insular cortex and indirect inputs--through the zona
incerta--from the cingulate and somatosensory cortex, and from
the central amygdaloid nucleus. Motor-related inputs reach the
parafascicular nucleus from the deep cerebellar nuclei and from
the motor cortex.

Inputs from the diencephalon and brainstem. Cornwall and
Phillipson67 discovered incoming fibers from the reticular
thalamic nucleus, the ventromedial hypothalamus, and the
mesencephalic reticular formation. Afferents from an area of the
pretectum which has been implicated in the detection of light

intensity77 are of special interest, considering that light is a
driving force in the behavioral assays of the present research.
The intralaminar nuclei, like most thalamic nuclei, receive
diffuse cholinergic, noradrenergic and serotonergic inputs from
the brainstem--the serotonergic component being particularly
heavy.78 The serotonergic link may allow indirect modulation of
the intralaminar nuclei by opioids, as the activity of the
serotonergic cells in the raphe dorsalis is increased by

Efferents to cortex, striatum and amygdala. A medio-lateral
organization of efferents from the center median and
parafascicular nucleus was discovered in fiber degeneration
experiments80,81 and labeling studies.54,82,83 In the monkey,
the lateral part of the center median nucleus sends projections to
the motor cortex, while the medial part is connected to the
sensorimotor striatum (the putamen, caudal to the anterior
commissure, and parts of the caudate; i.e., striatal regions with
inputs from the sensorimotor cortex). The high density of opioid
receptors in some of these striatal targets84 may suggest a role
of the parafascicular nucleus in opioid mediated functions.
The lateral part of the parafascicular nucleus is related to
motor regions, while the medial part predominantly interacts
with components of the limbic system.85 Ottersen and Ben-Ari86
demonstrated that the lateral part of the rodent and feline
parafascicular nucleus sends fibers to the ventral or limbic
striatum. Efferents to associative and limbic portions of the

striatum originate from separate neuronal populations. The
associative striatum includes the rostral putamen and the medial
part of the body and tail of caudate; i.e. regions with input from
the association cortex. The limbic striatum includes the nucleus
accumbens and the olfactory tubercle.54,82,83
The medial part of the parafascicular nucleus projects to
the central nucleus of the amygdala.86 The connections with the
amygdala are of particular interest, considering its role in
opioid-mediated endogenous pain inhibition87,88 and in the
acquisition and consolidation of conditioned fear.89 A direct
parafasciculo-cingulate projection has been reported by Musil and

Efferents to subthalamus. Feger et al.,91 in a retrograde
fluorescent double labeling study in the rat, found that distinct
parafascicular projections terminate in territories of the limbic
striatum and subthalamic nucleus which receive significant
prefrontal and cingulate inputs. Sadikot et al.54 investigated
efferents from the simian center median and parafascicular
nucleus to the subthalamus. The former projected to the
sensorimotor-, the latter to the limbic part of the subthalamus.

Efferents to thalamus and hypothalamus. Distinct
populations of neurons in the center median/parafascicular
complex were found to project to other thalamic nuclei, including
the mediodorsal-, midline-, reticular nuclei and zona incerta and
to the hypothalamus.54,63,83


The center median/parafascicular complex appears to be a
crossroads for sensory, affective and attentional aspects of
nociception, with access to basal ganglia systems that may form
a selector switch which chooses an appropriate motor
response.92 It is not fully understood to what degree the lateral
sensorimotor and the medial associative and limbic components
within the center median and parafascicular complex contribute
to the sensory, affective, motivational and motor responses to a
behavioral situation involving nociceptive stimulation. The
present study attempts to selectively disable the lateral
sensorimotor and the medial associative/limbic parafascicular
subsystems, in order to study the behavioral consequences for
animals trained to make adaptive responses to nociceptive
stimulation. With a particular interest in effects on nociception,
it is clear from the anatomical relationships of the intralaminar
nuclei that controls for effects on learning, motivation and motor
function must be incorporated into the experimental design.

Behavioral Data


The widespread connectivity of the center
median/parafascicular region suggests that complex
contingencies will determine the effects of lesions in this area
on behavior. This section reviews some of the relevant literature
in order to identify factors that will have an impact on the
behavioral consequences of lesions. The aspects to be discussed
include lesion effects on nociception and opioid action, the
importance of lesion size, lesion effects on response variability,
effects on memory, and implications of the structure of the
behavioral test.

Lesion Effect on Nociception and Opioid Action

Anderson and Mahan93 noted that cats, after undergoing
bilateral radiofrequency lesions in the center
median/parafascicular complex, exhibited elevated footshock
escape thresholds. This effect lasted 40-48 days and was
described as a recovering sensory deficit.
Yaksh and colleagues94 found that large lesions or local
anesthetic microinjections in the medial thalamus of rats failed
to alter tailflick or hotplate latencies. However, the lesions

potentiated the effect of morphine (6-20 mg/kg, IP) on these
behaviors. Likewise, narcotic depression of footshock escape
thresholds was potentiated by large medial thalamic lesions.95
Teitelbaum and coworkers96 were able to reverse opioid
tolerance--as inferred from EEG activity patterns--by destroying
the medial thalamus of rodent subjects.

Lesion Size

Effect on response means. Some authors have attempted to
make precise volumetric or planimetric estimates of lesion size.
Delacour and Borst97 pointed out that such estimates may create
a false impression of precision. Reports of lesion sizes as
percentages of the volume of the target structure must be
considered rough estimates, regardless of the numerical
precision reported, because anatomical boundaries are obscured
by the damage, and unpredictable distortion of the tissue may
result from the lesion.
Marburg98 reported that radiofrequency lesions had to cover
at least 40% of the center median/parafascicular region in order
to elevate the threshold for instrumental escape from electrical
footshock stimulation in monkeys. The magnitude of the effect
was only moderate. Mitchell and Kaelber99 trained cats to escape
from electrical tooth pulp stimulation by crossing a barrier. An
escape deficit was observed when > 50% of the center
median/parafascicular nucleus was destroyed. Interestingly, a

lesion of parts of the subparafascicular nucleus could substitute
for incomplete damage within the center median/parafascicular
nucleus --e.g., when the anterior or lateral portion of the latter
was spared. A similar size requirement existed when a foot shock
stimulus was used.100
Delacour26 studied the effect of electrolytic parafascicular
lesions on conditioned avoidance in rats. Rats were trained to
avoid an electrical footshock by crossing into the opposite
compartment of a two-way shuttle box when a light (conditional
stimulus) came on. A deficit in retention and reaquisition of
avoidance behavior was contingent upon destruction of at least
40% of the total volume of the nucleus.

Effect on response variability. Finger and Frommer101 used
a T-maze with differential floor temperatures to test the ability
of rats to discriminate temperature differences of 3, 7 and 12"C
within the range of 28 and 40'C. Bilateral electrolytic lesions in
the parafascicular region significantly raised the mean and
variability of the number of errors committed in the
discrimination task. The same paper reported that rats with
parafascicular lesions jumped more frequently than sham
operated controls when presented with electrical footshocks. The
variability of jumping frequency was larger in the lesioned
animals, which was attributed to a lesion-related decrease in the
reliability of a diffuse and redundant system.
Earlier it was pointed out that individual units within the
center median parafascicular region do not encode stimulus

intensity. However, the threshold differs from unit to unit, and
consequently the stimulus intensity may be encoded in the
integrated activity of a population of neurons. This concept of
statistical intensity coding predicts higher variability when the
population size decreases.

Lesion Effect on Memory

Fasciculus retroflexus. Electrolytic lesions but not
excitotoxic lesions in the parafascicular and mediodorsal
thalamic region compromised the performance of rats in a radial
maze experiment conducted by M'Harzi and coworkers.60 Subjects
could obtain a liquid reward only in an arm of the maze they had
not previously visited. It was a test of working memory, because
memory of the past response was required for correct execution
of the next response. The authors proposed that the fasciculus
retroflexus plays a role in working memory and is the critical
structure affected by the lesions. This view is inconsistent with
the findings of Brito et al.,102 who failed to observe working
memory deficits on a delayed alternation task after inflicting
collateral damage to the habenular nuclei and stria medullares
while placing electrolytic lesions in the mediodorsal nucleus of
rats. The stria medullaris, habenular nucleus and fasciculus
retroflexus are components of the same polysynaptic septo-
interpeducular pathway.57

Working memory deficits are more likely to interfere with
memory-intensive tasks, and consequently the performance in a
maze or a two-way shuttle box test is expected to be affected
more by a parafascicular lesion than the performance in a simple
task, such as a one-way shuttle box test.

Extinction rate and task retention. Roberts103 noted that
rats exhibited accelerated extinction of an aversive avoidance
task after a central gray lesion, but not when the lesion was
located in the parafascicular nucleus. Thompson104 reported that
rats with large bilateral pretectal lesions, which extended into
the parafascicular area and fasciculus retroflexus, suffered a
retention deficit for conditioned footshock avoidance when the
conditional stimulus was a light, but not when it was auditory.
Retention and relearning of the task were compromised severely
and were independent of the modality of the conditional stimulus
when the lesion was centered in the parafascicular region. In this
study there was poor anatomical definition of the lesions and
uncertainty regarding the relative contribution of sensory,
motivational and memory deficits, demonstrating the importance
of separate evaluation of each of these factors.

Role of Test Structure

Inhibition of innate responses. Delacour26 investigated the
effects of parafascicular lesions on the performance of rats on a

variety of behavioral tasks. A lesion-related deficit was observed
for a two-way footshock avoidance shuttle box test. General
motor impairment or a loss in motivational driving force were
ruled out as causative factors, because conditioned avoidance
behavior was not affected when tested in a one-way paradigm.
Vegetative responses (defecation, respiratory changes) to
footshock exposure were not diminished after bilateral
electrolytic lesions were placed, suggesting that the sensory
function was not compromised. However, the lesioned animals
frequently exhibited freezing behavior, which could interfere
with avoidance responding. The author proposed that the
parafascicular nucleus is critical for suppressing innate
responses, such as freezing, in favor of learned adaptive
behaviors. The probability of stereotypical behavioral responses
appeared to increase with the magnitude of the challenge the
animal faced. That is, the disinhibitory effect of the lesion on
innate behaviors became most visible in the more challenging
two-way test. Appetitive tasks were unaffected by
parafascicular lesions. One possible explanation is that the
probability of stereotypical behaviors interfering with the
instrumental response is low during appetitive tasks, reducing
the need for response conflict resolution. Alternatively, the
parafascicular nucleus may be specialized for aversive functions.

Innate behavioral responses, such as freezing or jumping, to
aversive brain stimulation103 or to electrical footshock101 were
not altered by lesions of the parafascicular nucleus in the
absence of an instrumental response option. Likewise, response

latencies in the tailflick- and hotplate test were unaffected by
even large lesions in the medial thalamus.94 However, Finger and
Frommer101 found that untrained rats with a parafascicular
lesion jumped with higher frequency in response to footshock
than control animals. The current threshold for jumping was not
affected by the lesion.
In summary, it appears that the parafascicular nucleus
allows the implementation of learned instrumental responses by
inhibiting competing unlearned behaviors that are likely to be
organized at a subcortical level. Interfering pressure from these
innate behaviors may be high in the presence of fear--e.g., when
aversive stimuli are expected. This model from previous research
predicts a lesion-induced behavioral deficit: (1) when the
response variable is a learned behavior, and (2) when conditions
are favorable for the occurrence of an innate response (e.g., under
high fear levels). No lesion effect can be expected when: (1) the
animal has not learned an instrumental response, or the test
conditions make adaptive response options unavailable, or (2)
innate response probability is low (e.g., when the driving force is
appetitive). We speculate that some types of instrumental
response are more sensitive to competitive interference from
innate behaviors than others. For example, freezing behavior may
interfere more with barrier crossing than with bar pressing.
Freezing and paw licking are not as likely to occur in a paradigm
developed by Vierck et al.,14 where the subject is restrained
with the forepaw resting on a response lever.

Escape tests vs. avoidance tests. Roberts103 elicited
vegetative behaviors (e.g., accelerated breathing, piloerection,
defecation) presumed to be indicative of aversion by electrical
stimulation within the nucleus gigantocellularis. The rodent
subjects were trained to escape from or avoid the aversive brain
stimulation by crossing between compartments in a shuttle box. A
comparison between the effects of electrolytic lesions in the
dorsal central gray and lesions in the parafascicular nucleus was
made: parafascicular lesions depressed avoidance more than
escape, while the reverse was true for central gray lesions.
Visceral and innate responses to the aversive stimulus were
reduced by central gray lesions but not by parafascicular lesions.
These findings are compatible with the hypothesis that innate
responses are inhibited by the parafascicular thalamus, if the
assumption is true that a conditioned avoidance task is more
susceptible to interference by innate behaviors than escape.

Clinical Data

Investigations in animal models have provided insight about
the involvement of the thalamus in short term nociception.
Studies in neurosurgical patients, on the other hand, have focused
on the role of this brain region in chronic pain. Feedback from
patients that underwent stimulation or lesions in the medial

thalamus is the only direct source of information about the role
of this region in conscious sensory experience.
Rinaldi and others105 recorded from thalamic single units
of patients with chronic deafferentiation pain and found irregular
spontaneous discharges in the lateral part of the mediodorsal
nucleus, the central lateral nucleus, the parafascicular nucleus
and a small part of the center median nucleus. In some of the
patients the spontaneous discharges made it impossible to record
evoked activity.
Electrical stimulation within the parafascicular nucleus of
a patient with intractable cancer pain produced an "unpleasant
feeling in the head", a feeling throughout her body "as if she was
going to explode" and a "pins and needles sensation" in one arm
and hand.106
Stereotaxic lesions (most commonly using radiofrequency
energy) have been placed in the somatosensory and medial
thalamus of patients with the objective of relieving chronic
intractable pain. Two main categories of chronic pain exist:
intractable nociceptive pain (e.g., cancerous invasion or arthritic
destruction of bone) and central pain. The latter can be caused by
peripheral nerve damage (deafferentiation, mechanical-, cancer-,
or herpes zoster-related nerve damage) or damage to CNS tissue
(spinal cord injury, or stroke).107
Case reports by Hecaen108 and by Urabe and Tsubokawa109
claim to have achieved temporary relief of nociceptive and
central pain by lesioning the center median nucleus. In some
cases the lesions had to be progressively enlarged over the course

of weeks or months, in order to prevent the pain from
Tasker107--in a review--indicated that medial
thalamotomy is more successful in relieving nociceptive pain
than central pain. Lesions in the medial thalamus may abolish the
intermittent shooting element of central pain but not the
constant burning or dysesthetic component. Initially stereotaxic
lesion efforts were directed against the center median nucleus.
Later the parafascicular and central lateral nuclei became
favored targets, because pain relief could be obtained with a
lower risk of concurrent sensory loss.
Recently, successful attempts have been made to place
thalamic lesions noninvasively with stereotaxically focused
gamma radiation.110,111 Steiner et. al.110 obtained the best
results when the center median/parafascicular lesions were
placed in close proximity to the third ventricle and posterior
commissure. Better relief was obtained for facial-, arm-, and
shoulder pain than for lower body pain. Unilateral lesions were
effective in some cases, especially when placed contralateral to
the pain source. Young et al.111 reported that lesions focused in
the medial thalamus reduced the attention of the patient to
chronic pain, but not the threshold to experimental phasic pain.


The combination of data from animal research and clinical
findings draw an incomplete picture of the functional role of the
parafascicular thalamic region. There is agreement that in most
instances a large lesion is required to produce an effect.
However, it is not clear whether large lesions are necessary
because they overcome redundancy within a diffuse system, or
because they are more likely to include critical sub-areas within
the parafascicular region. The medial part of the parafascicular
territory is of particular interest, firstly because some
clinicians found lesions to relieve pain best when placed
medially, and secondly because the fasciculus retroflexus, which
is suspected to be critical for performance deficits in some
behavioral assays, runs through the medial part of the nucleus.
Thus, it is necessary to better isolate the functional roles of the
medial and lateral portions of the parafascicular region.
The effects of lesions in the parafascicular region on
behavior were found to be dependent to a large extent on the
structure of the behavioral test. Some data suggest that the
nucleus does not have a sensory role, but contributes to response
selection, memory, aversive motivation or affective reactivity. A
reevaluation of the functional role of the parafascicular nucleus
with assays of pain sensitivity that can distinguish between
sensory, motivational, memory and motor effects is desirable.

The effects of parafascicular lesions on behavioral
responses elicited by phasic pain were generally modest.
Likewise lesions in this area of the thalamus in patients with
chronic pain had a limited effect on perception of phasic pain, but
often reduced the chronic pain. There is evidence from animal
research that the parafascicular nucleus modulates opioid effects
on thermal nociceptive reflexes. It is worthwhile to determine
whether such an interaction extends to opioid effects on
conscious thermal nociceptive behaviors. The functional role of
the parafascicular nucleus in pain could be limited to types of
pain that involve activation of endogenous opioid systems: tonic
or chronic pain is more likely to do so than phasic pain.

Research Goals

The rationale for planning the present study was based upon:

(1) Difficulties interpreting behavioral results of past
lesion studies, brought about by a lack of anatomical selectivity
of lesions targeted for the medial thalamus.
(2) The seemingly contradictory nature of past behavioral
results that may be attributable to critical differences in the
structure of the behavioral tasks that have been used (different
working memory requirements, different stress levels, different

types of stimulation). It is clear that no single behavioral
approach will provide a comprehensive answer regarding the
functions of the parafascicular nucleus. A definition of functions
for this nucleus will require integration of all positive and
negative results that have been obtained with different types of
(3) The possibility that the parafascicular region plays an
indirect role in conscious nociception by interacting with opioid

The main goals were the following:

(1) Employ a refined lesion methodology to selectively
destroy lateral or medial regions of the parafascicular nucleus.
Different functional consequences are expected to result from
such selective lesions, because the projection patterns of the
lateral and medial part of the nucleus are fundamentally
different. Different behavioral consequences are further expected
because medial lesions invariably lead to involvement of the
fasciculus retroflexus, while lesions limited to the lateral half
of the nucleus avoid this tract.
(2) Use two newly developed behavioral tests to assess the
effects of highly focused parafascicular lesions. The results will
add to the functional picture because the new behavioral assays
are structured differently from any test used previously to study
this region of the diencephalon.

(3) Study the effects of parafascicular lesions on the
influences of systemic morphine on operant behavior. Past lesion
studies have implicated the medial thalamus in regulation of
anti-nociception by morphine, but the lesions were not focused to
the parafascicular nucleus, and reflex-based nociceptive assays
were used.
(4) Contribute to the development of a functional model for
the parafascicular nucleus. Four alternate hypotheses that are
derived from the existing literature are proposed below.

Specific Aims

Find answers to the following questions that are concerned
with the anatomical distribution of parafascicular lesions:

(1) Do lesions limited to the medial or lateral part of the
parafascicular nucleus have differential effects on tendencies to
escape thermal stimulation, without producing motor deficits or
generally altering aversive motivation (for light-induced
(2) Does damage to the fasciculus retroflexus have an
effect on performance of escape responses?
(3) Does damage to the parafascicular nucleus or the
fasciculus retroflexus have an impact on retention of learned
escape behavior?

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EXSTA62W9_P1BDOK INGEST_TIME 2012-09-24T14:18:10Z PACKAGE AA00011829_00001