Brain Abnormalities in Murderers Indicated by Positron Emission Tomography.

Adrian Raine, Monte Buchsbaum, and Lori LaCasse

Chief Examiner Phil Banyard’s Introduction (abridged)

Are criminals any different to law abiding citizens?

In the 19^th Century it was thought that criminals had a certain definable appearance. In fact no difference has been found between ordinary people and criminals, using modern criminological research techniques.

Three basic techniques used to examine patterns of brain activity.

1. PET - Positron Emission Tomography - Used in the present study. Radioactivity is used to label blood, blood sugars or neurotransmitters (eg dopamine). Subjects are injected with the substance and then placed in a scanner that can detect what parts of the brain contain traces of the radioactive substance.
2. MRI - Magnetic Resonance Imaging - uses magnets and radio pulses to measure changes in oxygen levels (this indicates blood flows in the brain).
3. MEG - Magneto-Encephalography - uses very sensitive sensors to detect faint magnetic fields that are generated by active nerve networks.

Computers produce colour images of the brain. Problems and limitations with this type of evidence are:

1. The creation of the colour images go through many stages, at each stage errors may occur (technical errors or errors caused by wishful thinking by researchers.
2. Either excitory nerves or inhibitory nerves could cause a hotspot of brain activity. The scan can not tell the difference.
3. Well-learnt routines require less metabolism; So the hotspots may be showing what parts of the brain grapple with novel tasks.
4. MRI is susceptible to very snall movements; so a strong pulse in the forehead could contribute towards a false reading.
5. MEG is difficult to interpret and is greatly affected by magnetic fields in the general environment.

Abstract

Experimental group - 41 Murderers pleading Not Guilty by reason of Insanity (NGRI)

Controls - 41 matched for age and sex.

Used PET - Positron Emission Tomography.

Murderers had reduced glucose metabolism in

• Prefrontal Cortex,
• Superior Parietal Gyrus
• Left Angular Gyrus
• Corpus Callosum

Assymmetries (differences between the two hemispheres) found in

• Amygdala
• Thalamus
• Medial Temporal Lobe

Introduction

Other techniques have suggested generalised brain dysfunction (eg EEG, neurological, neuropsychological cognitive tests).

Prefrontal cortex - regulation of aggression, supported by neurological evidence from brain damaged patients.

Some studies have shown abnormalities in hemispheric asymmetries of function.

Reduced EEG interhemispheric coherence - trouble with the flow of information from one hemisphere to the next through the Corpus Collosum.

Recently reduced slow-wave amplitudes have pointed to left angular gyrus dysfunction.

Animals have been used to demonstrate the role of the Amygdala and Hippocampus in modulating aggression.

The Thalamus is the source of the hypothalamus-induced attack in cats.

Still not clear how violent offenders function.

Initial research using brain imaging has implicated frontal brain regions and the temporal cortex in brains of hospitalised aggressive patients. But a small sample was used and the patients had not committed seriously violent behaviour.

A pilot study of 22 offenders demonstrated prefrontal dysfunction.

The present study increases the sample to 41 and also includes an analysis of subcortical functions.

Hypothesis

A dysfunction will be found in areas previously implicated in aggressive patients and no dysfunction will occur in areas that are implicated in other psychiatric conditions not related to violence.

Methods

Subjects

·  ·  Murderers

39 men and 2 women from California with a mean age of 34.3 (SD = 10.1). Referred for evidence relating to their NGRI defence or incompetence to stand trial. They had many problems:

• Schizophrenia (6)
• Head injury or organic damage (23)
• Substance abuse (3)
• Affective disorder (2)
• Epilepsy (2)
• Hyperactivity and learning disability (3)
• Passive-aggressive or paranoid personality disorder (2)

·  ·  In 7 cases the nature of the crime suggested mental impairment. No drugs were administered 2 weeks prior to brain scanning. Urine screens verified this.

Controls

Matched on sex and age.

The 6 schizophrenics were matched with control schizophrenics of the same age and sex.

Mean age 31.7 (SD= 10.3) (No significant difference between the ages of the 2 samples).

Controls were screened for health and their medical history and a psychiatric interview was given.

They were receiving no medication, were healthy and had no history of psychiatric illness.

Potential subjects with a history of seizure disorders, head trauma or substance abuse were excluded.

The study was subject to the ethical guidelines of the University of California.

PET Task Procedure

Fluorodeoxyglucose (FDG) was used as the tracer. There was a 32-minute uptake before the Continuous Performance Task (CPT). Split half reliability was high on this task (r= .843, p<.001).

10 minutes before the FDG injection practice was given on the CPT.

To avoid the novelty of the CPT being FDG labelled the CPT was started 30 seconds before the FDG injection was administered.

After the 32 minutes the subjects were then placed in the scanner, which was in an adjacent room.

The subjects were scanned at 1cm intervals laterally, and ten slices were taken.

Two techniques were used. One for lateral areas and one for medial areas.

Cortical Peel technique (lateral areas).

Glucose values were recorded as relative to other values within the slice. The advantages of this were:

1. Removes whole brain metabolic rate
2. Relative values are more widely reported
3. More likely to be related to function in specific neuroanatomical systems
4. Shows greater reliability

The areas studied are shown below.

Box technique (medial areas)

Medial cortical and subcortical regions studied (inside the brain)

3x3 pixel region of interest

A pixel is 2mm2

The areas studied are shown below

Summary of areas studied

Cortical areas

• Lateral prefrontal
• Medial prefrontal
• Parietal
• Occipital
• Temporal
• Cingulate

Subcortical structures

• Corpus callosum
• Amygdala
• Medial temporal lobe and hippocampus
• Thalamus
• Putamen
• Globus pallidus
• Midbrain
• Cerebellum

Results

Structure	Level of significance comparing murderers to controls (2 tailed)	Higher or Lower metabolism. Murderers compared with controls	Notes	Findings from previous research
Prefrontal - Lateral	p<0.02	Lower		Prefrontal Deficits Impulsivity, loss of self-control, immaturity, altered emotionality, inability to modify behaviour.
Prefrontal - Medial	p<0.02	Lower
Left Angular Gyrus	p<0.06 or p<0.03 (1 tailed) (see note)	Lower	Could justify a one-tailed hypothesis as the left angular gyrus is associated with verbal ability, and criminals tend to have a lower verbal IQ	Posterior parietal cortex (including superior and angular gyri) is involved in the integration of sensory input and the formation of abstract concepts. The reduction in glucose metabolism in the left angular gyrus correlates with reduced verbal ability. Damage results in deficits in reading and arithmetic. Does failure in education lead to crime? Violent offenders tend to have low verbal IQs.
Temporal	NS
Occipital	P<0.02	Higher
Occipital area 17	P<0.0001	Higher
Occipital area 18	P<0.001	Higher
Corpus Callosum	P<0.001	Lower	Brain damaged murderers had lower metabolism compared with other murderers (p<0.08)	This connection between the two hemispheres could contribute to the asymmetries of function. Reduced interhemisperic integration has been observed in antisocial and violent groups. The right hemisphere can produce a negative affect. If the connections between the two hemispheres are poor then the left hemisphere is unable to positively influence the negative thinking of the right. Rats that are right hemisphere dominant for mice killing, kill more when the corpus callosum is cut. Split brain patients (see Sperry) often display inappropriate emotional expressions. They also are not able to grasp the long-term implications of a situation. Violent offenders also have this problem.
Amygdala	P<0.02 group x hemisphere interaction	Lower on the left and Higher on the right	Left handed murderers, compared with right handed murderers, had significantly less asymmetry (i.e. less difference between the right and left hemisphere amygdala activity) (p<0.002)	Limbic deficits Aggressive behaviour in animals and humans. Amygdala, hippocampus and prefrontal cortex form part of the Limbic System, which is associated with the expression of emotion. The recognition of affective and socially significant stimuli. Socially inappropriate behaviour has been shown by some violent individuals (that is in addition to their violent behaviour!). Destruction in animals results in lack of fear. Destruction in man brings about a reduction in autonomic arousal. The amygdala functions in parallel with the object recognition system of the posterior parietal cortex. Misappraisal of ambiguous stimuli in social situations can lead to violence.
Medial temporal lobe including the hippocampus	P<0.005 group x hemisphere interaction	Lower on the left and Higher on the right		Hippocampal formation is associated with aggression in cats. With the septal area and prefrontal cortex it forms a behavioural inhibition system The Hippocampus, Amygdala and Thalamus are responsible for learning, memory, attention. It could be that criminals may not learn from experience.
Thalamus	P<0.04 group x hemisphere interaction	Same on the left and Higher on the right		Relays inputs from subcortical limbic structures to the prefrontal cortex.
Cingulate	NS
Caudate, Putamen, Globus Pallidus, Midbrain and Cerebellum	NS		These areas have never been associated with aggression, but have characterised other mental disorders.

Behavioural Performance on the CPT

No significant difference

Effect of Handedness, Head injury and Ethnicity

Although subjects were matched on gender, age and schizophrenia, it was not possible to match them on handedness, head injury and ethnicity.

6 left handed murderers were compared to 6 right-handed murderers. Results were significant for the amygdala (see note in above table) and no where else. Remember there were only 6 subjects and that it is difficult to detect significance with such a small number of subjects. It is concluded that left handedness does not explain the differences reported in the above table.

14 non-white murderers were compared with white murderers - no significant difference.

23 murderers with a history of head injury were compared with murderers without a history of head injury. No significant differences were found except there was a tendency for brain damaged patients to have lower activity in the corpus callosum (p<0.08).

Discussion

The data confirms deficits in the prefrontal cortex from earlier pilot study, plus it yields new findings.

Biosocial pathways from Brain Deficits to Violence

Here the present findings are related to the findings of previous research. I have incorporated some of these points in the last column of the above table.

Anterior parietal regions are responsible for basic somatic sensations and perceptions. There was no difference in functioning between the murderers and the controls.

Conclusion

Many different areas of the brain contribute to violence in the presence of other social, environmental and psychological predisposition.

Not all brain areas related to aggression were studied (e.g. septum and hypothalamus). We therefore have not got a complete picture, but murderers pleading NGRI have different brain functioning compared to normals.

Potential Confounds

The results are unlikely to be owing to chance because:

1. Large sample size
2. Strong effects
3. The areas studied were selected on the basis of prior theorising; all but one of these areas produced significant results.
4. The MANOVA statistical test was used; this takes into account more than one variable at a time. Also two-tailed tests were used. This reduces the risk of a type 1 error (accepting the experimental hypothesis in error).
5. Other brain areas that were unrelated to violence did not yield significant differences. This suggests a double dissociation between the areas related to aggression and areas associated not associated with violence.

·  Some effects are marginal (e.g. left angular gyrus) and some were not predicted a priori (therefore could be owing to chance) (e.g. subcortical laterality effects and increased occipital functioning).

No IQ scores were obtained.

Lower IQ results in higher glucose metabolism, so low IQ can not explain low glucose metabolism found in murderers (e.g. lateral prefrontal, corpus callosum, medial prefrontal, parietal cortex).

Specificity of Findings

Could the murderer’s schizophrenia have accounted for the findings?

Psychiatric patients show abnormalities that are not found in murderers and vice versa.

Schizophrenics have altered functioning in the:

• lateral temporal cortex
• basal ganglia
• cingulate gyrus
• caudate
• cerebellum.

These structures are unaffected in murderers.

Depressives have dysfunction lateralised to the left hemisphere (particularly the left dorsolateral prefrontal region). Whereas the murderers show bilateral reduction in glucose metabolism. In depressives the caudate nucleus and the cingulate gyrus are affected, whereas these are unaffected in murderers.

Obsessive-compulsives show higher, not lower, glucose levels in the orbitofrontal cortex.

Cannabis use affects cerebullar functioning whereas in murderers this is normal.

Detoxified alcoholics show increased (not decreased) brain metabolism during detoxification. The basal ganglia show low metabolic rates, whereas this is unaffected in murderers.

Chronic alcoholics show low cerebellar metabolism, whereas the metabolism was normal in muderers.

The reduced prefrontal activity, found in this study, is not specific to violence; it is also observed in many different psychiatric conditions.

The left lower than right asymmetries in the Amygdala, Thalamus and Hippocampus, coupled with the dysfunction of the corpus callosum and the left angular gyrus, seem specific to murderers.

Problems

1. Because of differences in imaging techniques it may not be possible to make strict comparisons.
2. Murderers pleading NGRI do not constitute a homogenous clinical group. There would be large differences in behaviour within the group.

Strengths, limitations and conclusions

1. PET scanning has limited accuracy in locating specific structures.
2. Many of the psychiatric studies, with which the present study was compared, lacked control groups.
3. NGRI murderers are a select subgroup - results may not generalise to violence per se.
4. Some findings (subcortical asymmetries and the occipital cortex) are not predicted and therefore need to be replicated.

 

Strengths

1. Large sample
2. Seriously violent sample
3. Matched for age, sex and schizophrenia
4. Confounds ruled out: handedness, ethnicity and head injury.
5. Both groups were equally good at the CPT (task).
6. Medication controlled.

 

We can not conclude

1. Violence is determined by biology alone.
2. Murderers pleading NGRI are not responsible for their actions.
3. Cause nor causal direction has been established.
4. It is possible to generalise to other types of violent offenders
5. Specificity to violence has been established (This could be established by comparing the results of the murderers with a non-violent criminal control group).

Phil Banyard’s Discussion (abridged) 

Further limitations

1. Imaging techniques are still in their developmental stage and should therefore be treated with caution.
2. The task used has nothing to do with violence.
3. The level of violence was not controlled. NGRI’s could have murdered somebody with poison, a gun, by strangulation or with a chainsaw. All of these various methods of dispatching a victim involve different degrees of violence.

It is difficult to generalise from these results. We would certainly not wish society to perform brain surgery or administer strong drugs to these individuals who have been found to have brain functioning patterns that are similar to those found in the murderers in this study. We certainly could not be confident that we could identify those who are predisposed to committing violent crimes.