IUBio

the neural basis of fear

Erlick Pereira eacp2 at hermes.cam.ac.uk
Fri May 11 17:05:43 EST 2001


Only in the last two decades has psychology focused upon the coordination of
behavioural responses to fear.  Prior to then, fear was considered merely a
type of emotion and emotions had received little attention in deference to
cognition and MacLean's (1951) conceptualisation of a catch-all subcortical
"limbic system".  MacLean's "visceral brain" could preconsciously modulate
somatic responses to emotion-eliciting stimuli, and influence cognition and
conscious perception to precipitate affective disorders of anxiety and mood.
MacLean's functionalisation of Papez's (1937) limbic system has been
criticised on several fronts, most notably that it seems to suggest that any
structure involved in emotional processing belongs to the limbic system.
The concept therefore becomes so expansive as to be irrefutable, which is
hardly surprising given its neural correlation to such a vague and
ill-defined term as 'emotion'.  It seemed clear that scientific progress
required focus upon an operationally defined and experimentally tractable
facet of emotion.  As LeDoux (1996) explains, fear provides such a paradigm,
measurable in terms of objective somatic and behavioural parameters.
Fearful responses can be classified into:
1. autonomic, e.g. heart rate (HR), blood pressure (BP), galvanic skin
response (GSR)
2. endocrine, e.g. enhanced release of the adrenomedullary hormone
adrenaline or adrenocortical corticosteroids via potentiation of the
hypothalamo-pituitary-adrenal axis
3. reflexive behavioural, e.g. height jumped when startled, freezing,
blinking
4. complex behavioural, e.g. number of conditioned lever presses suppressed
or avoided with a fearful stimulus such as electric shock.
Measurement of objective thresholds, rather than subjective feelings, also
enabled the application of more invasive and controllable experimental
methods upon rodents and primates, rather than humans in whom the spatial
and temporal resolution of neural localisation methodologies is somewhat
cruder.  Finally, fearful expression is a highly phylogenetically preserved
phenomenon, being observed in flies, worms and snails as well as reptiles,
birds, mammals and, of course, humans.  Cross-species homology from rodent
to man is therefore robust, whereas it may not necessarily be so for other
emotions such as anger.
 In this essay, I shall argue that animal experiments, particularly in
rodents, involving lesions and pharmacological manipulations, suggest that
the discrete nuclei of the amygdala have distinct functions in processing
and generating responses to fearful situations.  Experiments elucidating
thalamic input and neocortical modulation enable thalamo-amygdale-cortical
'fear circuits' (LeDoux, 2000) to be proposed for mediating fear response
behaviours.  Furthermore, amygdala interaction with hypothalamic and
brainstem circuits (Siegel, 1999) reveals potential mechanisms by which
autonomic and endocrine responses to fear might be coordinated.  Recent
neuroimaging and clinical studies augment this evidence.
The paradigms focused upon utilise unconditioned electric shocks,
classically conditioned auditory tones and instrumentally conditioned lever
presses or avoidance to assess animals' detection and processing of
dangerous, hence fear-eliciting, stimuli.  Other paradigms that lie beyond
the scope of this essay, yet elucidate how other emotions are coordinated,
include appetitive stimulus-reward association learning (Everitt and
Robbins, 1992, Gallagher, 1994, Rolls, 1999) and Gray's (1982) theory of a
septo-hippocampal behavioural inhibition system, suppressed in anxiety.
Most researchers, notably Davis and LeDoux, have focused upon the amygdala
as the neural locus of fear mediation and response control.  The amygdala is
a complex structure of at least twelve discrete subregions.  The nuclei
focused upon here are lateral (LA), basolateral or basal (BL), basomedial or
accessory basal (BM) and central (CE).  Projections between nuclei are
depicted below, but it should be noted that alongside inter-nuclei
projections there are numerous intra-nuclei and inter-subnuclei projections
representing stimuli in parallel with different consequences such as memory
or homeostatic control (Pitkanen, 1997).  Considering the nuclei below
offers the simplest complete analysis of the neural coordination of fear.
 Pavlovian fear conditioning using auditory tones and place, together with
excitotoxic lesioning, has revealed place conditioning to require ventral
hippocampal integrity (CA1, subiculum) which is consistent with its well
documented role in spatial cognition (O'Keefe and Nadel, 1978), declarative
memory for contextual, configural associations (Squire, 1992, Eichenbaum,
1996) and scenes (Gaffan, 1992).  Tract tracing reveals hippocampal
projection to BL, suggesting it may imbue context upon amygdale fear
association and processing.  LA nuclei are found by single cell recording
(SCR) to have a short latency response to auditory tone coupled with
electric shocks, acquired rapidly within one to three trials (Quirk 1995,
1997).  Thus, LA is believed to be the initial site of conditioned stimulus
(CS)-US (unconditioned stimulus) convergence and association.  LA lesions
are found to impair acquisition of the rat freezing response to the tone,
thus implicating this nucleus in acquired fear.  BM and CE neuronal
plasticity to tone-shock pairings is also observed, but these are of longer
latency and take longer to acquire.  CE has traced output to the brainstem
(Davis, 1992) and its lesion inhibits conditioned fear expression without
affecting LA neuron responses.  Thus, a projection of CSàLAàCE is proposed.
Selective lesions of brainstem sites eradicate specific autonomic
expressions of fear, e.g. lateral hypothalamus and blood pressure, or
periaqueductal grey and freezing (LeDoux, 1988).  Similarly, lesion of the
bed nucleus of the stria terminalis, which has strong innervation from CE,
inhibits hypothalamo-pituitary-adrenal axis endocrine responses to fear.
Considering unconditioned stimuli, LA has strong innervation from the
thalamus and nociceptive afferent innervation from spinothalamic tracts
(LeDoux, 1990).  The thalamus relays integrative input from all sensory
modalities to the cortex, as well as cortical reciprocal modulation.  Thus,
LA seems anatomically optimal for associating sensation, be it visual,
auditory, olfactory or tactile, with pain.  Projections from sensory,
association and frontal cortices also converge upon LA, which may therefore
process unconditioned stimuli by either pathway.  LeDoux (1996) has proposed
that thalamic input is "quick and dirty" versus slow and detailed cortical
input, the latter modifying the former and reflecting different phases of
fear initiation and expression.  The LA nucleus therefore seems the supreme
associator and integrator of sensory and aversive stimuli projecting
LAàBL/BMàCE.  BM also receives posterior thalamic input (LeDoux, 1990)
therefore may utilise this sensory information to modify hippocampal input
conveying contextual CS.  CE receives its own nociceptive spinal cord input,
suggesting modification and integration at each progressive level of a
intra-amygdaloid 'lattice hierarchy' from LA input to CE output (Gallistel).
However, Kilcross, Everitt and Robbins (1997) observed that CE lesion
prevented conditioned suppression and BL lesion prevented avoidance
behaviour in a concurrent task whereby rats could suppress responding upon
an aversively coupled lever or avoid responding by pressing a neutral lever.
They proposed that BL is therefore necessary for expression of voluntary,
instrumental avoidance behaviour, possibly acting via projections to ventral
striatum and orbitofrontal cortex modulating affective and motoric behaviour
(Taylor, 1986, Alexander, 1990, Damasio, 1994), whereas CE is necessary for
expression of automatic, Pavlovian responses, possibly acting via brainstem
projections.  The suggestion that CE can directly mediate fear responses
alone contradicts a hierarchical model.  However, Kilcross's experiments
involved autoshaping over many trials (~120) whereas fear responses over
several trials are likely to require LA nuclei.  Thus, LA remains necessary
for rapid acquisition of fear responses.  The suggestion that BL moderates
its own instrumental expression pathways is not inconsistent with a
hierarchical model, albeit somewhat less parsimonious.  Its output may be
modified by reciprocal CE input.
Long-term potentiation (LTP) is a phenomenon first observed in the
hippocampus (Bliss and Lomo, 1973) as a neural substrate of long-term
associative, hence Hebbian and Rescorla-Wagner, learning postulated to
underlie memory.  Studies of LTP in the amygdala in vitro and with
N-methyl-D-aspartate (NMDA) channel blocking drugs such as AP5 suggest that
LTP mirrors fear conditioning acquisition.  AP5 prior to training, not
testing, blocks conditioned fear acquisition and expression, although the
two may be affected independently (Davis, 1997), perhaps by blocking LA and
CE respectively.  It seems that experimental animal studies implicate the
amygdala as modulating behavioural, autonomic and endocrine responses to
fear and associating previously neutral stimuli with fearful responses in
Pavlovian and instrumental contingencies.  The LA and CE nuclei appear to be
the sites of predominant input and output, respectively to and from
brainstem, striatum, hippocampus and cortex.  Amygdala activation is
greatest at the early stages of acquisition and LTP maybe a mechanism
involved in preserving these long-term associations.
Considering human studies, recognition of facial expressions, particularly
fearful faces is inhibited with amygdala damage (Adolphs, 1995, Calder,
1996).  Positron emission tomography (PET) in normals shows normals to have
enhanced amygdala activity for angry and fearful faces (Breiter, 1996), both
consciously and subconsciously (Morris, 1998) and functional magnetic
resonance imaging (fMRI) shows similar enhancement with fear conditioning
(LaBar, 1998).  Cross-correlations with thalamic and collicular activation
support propositions of sensory input from these regions to the amygdala
(Morris, 99).  However, whilst the amygdala seems central to fear
processing, it cannot alone elicit all the responses described.  At this
level of the neuraxis, other centres are invariably involved modulating
response expression.  Clinical studies of pathologies elucidate this axiom.
Post-traumatic stress disorder, involving the recurrent activation of
traumatic memories and fearful responses, is associated with hippocampal,
prefrontal cortex, amygdala and stress hormone abnormalities (Pitman, 1999).
Panic disorder and phobias also elicit fear, yet these three disorders are
not easily extinguishable, as they would be if merely conditioned by the
amygdala.  Seligman's (1971) preparedness theory that some stimuli, such as
snakes, are biologically of greater adaptive benefit to be less
extinguishable than others, provides a cognitive explanation, but neural
explanation requires animal studies.  Medial prefrontal cortex lesions in
rats (Morgan, 1993) are found to diminish extinction of conditioned fear,
suggesting that perturbation of the supervisory attentional role of the
prefrontal cortex (Norman and Shallice, 1986) in cognitively modulating the
processing of fear may be responsible.  The amygdala both directly and
indirectly projects to prefrontal cortex, directly via reciprocal
innervation and indirectly via modulation of midbrain monoaminergic sites
that innervate and modulate prefrontal cortex function.  Furthermore, recent
SCR has shown the amygdala to be necessary in the prefrontal cortex for the
formation of representations of the degree of predictability of aversive
events (Garcia, 1999).
To summarise these complex neural interactions, my thesis is that
fear-eliciting stimuli, either unconditioned or conditioned reach the
amygdala, predominantly the LA nucleus, first by fast thalamic and
nociceptive input.  The amygdala then effects the following responses,
predominantly from the CE nucleus:
1. brainstem sympathetic nervous stimulation causing various somatic
effects, notably adrenaline secretion from the adrenal medulla, leading to
increased chronotropic and inotropic cardiovascular effects
2. stria terminalis neural stimulation of hypothalamus corticosteroid
releasing factors which perfuse the anterior pituitary causing the humoural
release of ACTH hormone which enhances corticosteroid release from the
adrenal cortex
3. midbrain stimulated release of modulatory neurotransmitters including
dopamine,  noradrenaline, and acetyl choline that have complex interacting
effects subcortically and cortically.  To describe them in very superficial
terms, dopamine from ventral tegmental areas serves to enhance the salience
of stimuli, noradrenaline from locus coeruleus enhances the signal to
response ratio to salient stimuli and attenuates neural responding to
non-salient stimuli, acetyl choline from basal nuclei enhances cognition and
memory.  Peripheral adrenaline also stimulates locus coeruleus noradrenaline
release via vagal afferents and the overall arousing effect of the
monoamines is concentration dependent in an inverted-U, Yerkes-Dodson
fashion, although it is complicated by their interactions.
4. striatal projections may enhance or inhibit initiation of motoric
responses (fear-potentiated startle, freezing) as well as modulating the
affective and cognitive components of stimuli by feeding into
cortico-striatal loops (Alexander, 1990)
5. Prefrontal cortex projections may enhance the immediate recall of a
fearful event, alter the preparation of responses to it and the attention
paid to it (after Fuster's thesis upon prefrontal cortex functions, 1998).
In particular, orbitofrontal cortex projections may act to enforce "somatic
markers" governing personality, impulses and "the feeling of what happens"
(Damasio, 1994, 1999)
6. Hippocampal connections may enhance the emotive component hence salience
of declarative memories associating temporal and spatial contexts with
events.
Each of the systems mentioned above can be deconstructed in its own right
for its role in processing of fear as can other neurochemicals like GABA and
 serotonin, but central to all are the initial associations between stimuli
made in the amygdala.  Such associations affect, and to a lesser extent are
modulated by, attention (prefrontal cortex, parietal cortex, anterior
cingulate cortex), memory (hippocampus, cortex) and consciousness (enhancing
the perpetuation of distributed synchronous binding neural oscillations,
perhaps? (Singer, 1999, Crick and Koch, 1990)).  The amygdala clearly plays
a crucial operational role at the interface of emotion and cognition, and
fear coordination is at present its most experimentally tractable facet.

Opinions?





More information about the Neur-sci mailing list

Send comments to us at biosci-help [At] net.bio.net