News & Views | Published:


Fear thou not

Mice lacking a certain neurotransmitter receptor have trouble forgetting scary experiences. This finding uncovers a fear-regulating feedback loop in the brain that might be at work in humans, too.

“Show me a man,” said the Roman writer Seneca, “who is not a slave; one is slave to lust, another to greed, another to ambition, and all men are slaves to fear.” There are few who would seriously challenge this 2,000-year-old conclusion, and fear of fear is probably one of the major motivations for studying emotions in general. Until quite recently, however, it was considered unfashionable for respectable brain scientists to investigate the molecular and cellular bases of the emotions. This has now changed, and we have ample information about the brain circuits that encode fear1. Writing last month in Cell, Shumyatsky and colleagues2 have added important details to the emerging picture. Besides exposing tricks used by the brain to manage fear, the new findings might, in the long run, pave the way to improved treatments for pathological fear responses such as post-traumatic stress disorder.

Study of the cellular and molecular bases of fear requires a model system in laboratory animals. The most popular model is classical (alias pavlovian) 'fear conditioning'. To understand how this works, it helps to recall how Pavlov trained his dogs in his St Petersburg lab a century ago. In a typical experiment, a dog was presented with a sound, and immediately afterwards with meat, which evoked salivation. In the language of psychology, the sound is the conditioned stimulus, the meat the unconditioned stimulus, and salivation in response to meat is the unconditioned response. With time, the dog learned that the sound predicted food, and salivated when presented with the sound alone (the conditioned response). Now, let's substitute the dog with a mouse (or rat), keep the sound, but replace the food with an electric shock to the foot. The result is that the mouse learns to fear the sound. Fear has many manifestations, one of which — freezing of movement — is commonly used as the conditioned response in lab animals.

Part of the neuronal circuitry that subserves fear conditioning is found in the amygdala, a collection of neural structures in the temporal lobe of the brain that controls many aspects of emotional and social behaviour. According to a generally accepted model, information about the conditioned and unconditioned stimuli becomes associated in the lateral nucleus of the amygdala, and output from the amygdala controls the expression of fear (reviewed in ref. 3). Although its exact role is still uncertain4, the amygdala is known to be essential for the formation of memory of fearful experiences. Furthermore, in keeping with the conceptual framework that underlies contemporary neurobiology, it is believed that 'fear' in the amygdala to previously innocent stimuli is caused by experience-induced changes in its synapses (the connections between nerve cells).

The molecular basis of fear conditioning is one of the hottest topics in memory research5,6,7, and Shumyatsky et al.2 have now tackled certain aspects of it. First, they compared gene expression in different brain areas in mice. This led them to identify a gene coding for a neurotransmitter (a molecule that transmits information between neurons) called gastrin-releasing peptide (GRP), which they found to be preferentially expressed in the principal type of nerve cells in the lateral amygdala and in some interconnected brain areas. The authors then searched for cells that contained the GRP receptor (GRPR) and that could therefore react to GRP. They found the receptor in a subpopulation of amygdalar interneurons — another type of nerve cell — that release the neurotransmitter γ-aminobutyric acid (GABA).

GABA inhibits nerve-cell activity, and Shumyatsky et al. found that GRP causes the GRPR-expressing population of interneurons to release more GABA, augmenting their inhibition of the principal cells (Fig. 1a). This effect was abolished in mutant mice in which the GRPR gene was knocked out (Fig. 1b). Moreover, these mutants also displayed enhanced long-term potentiation (LTP) in the neuronal pathway linking the cortex — which perceives and processes fear-associated stimuli, such as the sound in fear conditioning — and the amygdala. LTP is an enhancement of synaptic transmission that persists after repetitive stimulation of a particular neuronal pathway, and it is a major model for learning-related changes in the mammalian brain. Moreover, LTP has been shown to contribute to fear learning in vivo8. So, this all adds up: an absence of GRPR impairs the inhibition of the principal amygdalar cells by interneurons, and this allows synapses between the neurons of the cortex and amygdala to become more susceptible to LTP.

Figure 1: Neuronal circuits and frightened mice.

Shumyatsky et al.2 have proposed the following model for a negative-feedback loop that modulates the activity of the principal neurons in the brain's lateral amygdala, thereby controlling memories of fearful stimuli. a, In normal mice, the principal neurons process and transmit information relevant to the formation of fear-associated memories. On activation, these neurons release (among other neurotransmitters) the peptide GRP onto inhibitory interneurons that carry the GRP receptor (GRPR). These interneurons release the inhibitory transmitter GABA onto the principal neurons. Activation of the GRPR stimulates the interneurons to release more GABA, increasing inhibition of the principal neurons. b, In mice in which the GRPR is knocked out, the inhibitory interneurons are insensitive to GRP, the feedback loop is severed, and the principal neurons are not properly inhibited. The mice form an abnormally strong and persistent fear memory. The innate level of anxiety and memories of non-fearful events are not affected. (Modified from ref. 2.)

Together, the data indicate that the lateral amygdala contains a loop that involves the principal neurons and the GRPR-containing inhibitory interneurons. These interneurons regulate the fear-related information that the principal cells process. When the principal neurons fire, they release, among other molecules, GRP, which interacts with GRPR on the interneurons and enhances their inhibition of the principal neurons. So this is a negative-feedback loop.

How is this relevant to behaviour? Shumyatsky et al. found that the GRPR-deficient mice displayed a greater and more persistent long-term memory of fear. These mice were not, however, generally anxious, nor did they have a stronger memory of non-fearful events. In other words, the GRP-mediated negative-feedback loop in the amygdala seems to be involved specifically in registering emotionally traumatic experiences.

Several points are noteworthy. First, this work epitomizes the kind of multi-level approach that is essential for real progress in memory research. Second, Shumyatsky et al. have dissociated two tiers in a mechanism that regulates the balance between excitation and inhibition in the amygdala. One tier, involving the basal activity of the inhibitory neurotransmitter GABA, controls the innate level of anxiety; the GRPR mutants might have lived happily for ever in a world without painful surprises, because normal basal levels of GABA are released from the inhibitory interneurons in their amygdalas.

The other tier involves the fine-tuning by experience of this balance between excitation and inhibition. This tier constrains the reaction of organisms to emotional trauma. Here GRP is important, and so mutants with a defective GRP-mediated negative-feedback loop overreact to fearful life experiences and have trouble forgetting them, presumably because they lack the mechanism that, in normal mice, quenches fear-induced overactivation of the principal amygdalar neurons. A final point is that only a subpopulation of amygdalar interneurons responds to GRP; other amygdalar circuits, using different neurotransmitters, might fine-tune other emotions.

It will be interesting to see whether these findings translate to humans. Might, for instance, experience-dependent alterations in the GRP-mediated feedback loop contribute to post-traumatic stress disorder following devastating experiences such as rape or a terrorist attack? Might defects in this neuronal system be identified in people who are particularly susceptible to post-traumatic stress disorder? And do some of us overexpress GRP or GRPR, ultimately rendering us less easily frightened than others because we forget scary experiences more quickly? Surely, we are all anxious to know the answers to these questions. But even before we do, the GRP system is likely to become a focus for research into new, selective, anxiety-relieving drugs.


  1. 1

    LeDoux, J. E. The Emotional Brain: The Mysterious Underpinning of Emotional Life (Simon & Schuster, New York, 1996).

  2. 2

    Shumyatsky, G. P. et al. Cell 111, 905–918 (2002).

  3. 3

    Aggleton, J. P. (ed.) The Amygdala: A Functional Analysis (Oxford Univ. Press, Oxford, 2000).

  4. 4

    Cahill, L., Weinberger, N. M., Roozendaal, B. & McGaugh, J. L. Neuron 23, 227–228 (1999).

  5. 5

    Maren, S., Aharonov, G., Stote, D. L. & Fanselow, M. S. Behav. Neurosci. 110, 1365–1374 (1996).

  6. 6

    Lamprecht, R., Farb, C. R. & LeDoux, J. E. Neuron 36, 727–738 (2002).

  7. 7

    Ressler, K. J., Paschall, G., Zhou, X. L. & Davis, M. J. Neurosci. 22, 7892–7902 (2002).

  8. 8

    Rogan, M. T., Staubli, U. V. & LeDoux, J. E. Nature 390, 604–607 (1997).

Download references

Author information

Correspondence to Yadin Dudai.

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.