From the archives...
It probably depends on how you define empathy. Empathy, by any definition, implies emotional sensitivity to the affective state of another. Sometimes the empathy response is automatic or reflexive, like when babies start to cry upon hearing another baby crying. Sometimes a strong cognitive component is required, such as for compassion. A more specific understanding of empathy requires similarity between the affective states of the observer and the observed, with an understanding that the observer feels a certain way because he has observed the affective state of the other.
One signal that humans and other primates use to perceive fear in others is facial expression. In mice, fear is usually indicated by freezing and becoming immobile. This doesn't seem like a particularly smart survival strategy to me, because if a hawk is coming down to get you, wouldn't it be better to RUN AWAY? But nonetheless, mice freeze when they are afraid.
Jeon and colleagues used this fact to investigate whether some mice - observing expressions of fear in other mice - would likewise show fear responses? If they did, what was the neural substrate?
In classical conditioning, an animal (the mouse, in this case) is put into a cage and given electrical shocks. The mouse will learn to associate the particular cage with the painful shocks, and later, when placed back into that cage, he will display a fear response (freezing), even if he isn't given any shocks. In this procedure, they had a second mouse (we'll call it "the observer") in an adjacent cage observe the conditioned mouse (we'll call it "the demonstrator"). Sometimes the demonstration and observation cages were separated by a clear plexiglass partition, and sometimes they were separated by a partition made of opaque black plastic.
It turns out that even though the observer wasn't being conditioned at all, and never received any electrical shocks, it did display a fear response. The response was context-specific - it was only displayed when the mouse was placed back into the same observation cage.
The effect is strongest when the observer could visually see the demonstrator. However, there was a slight effect even when the partition was opaque - this is probably due to auditory or olfactory (sound- or scent-related) cues.
This demonstrates that mice can learn fear through observing other mice who are experiencing fear. And not only that, but the social relationship between the observer and demonstrator mice mattered! When the demonstrator and observer mice were siblings, the effect was more pronounced than for non-sibling pairs:
When the demonstrator and observer mice were mating partners (and lived in the same home cage), there was a graded effect with those living together the longest showing the strongest effects!
These results mean that the social relationship between observer and demonstrator is central to the formation of an empathic fear response in the observer mouse! (But wait, there's more!)
What was the neural substrate responsible for the empathic fear response displayed by the observer mouse? The authors identified several candidate structures based on what is known to be involved in fear conditioning:
(1) The lateral nucleus of the amygdala is essential for fear conditioning
(2) Part of the cortical somatosensory area, as well as the ventral posterolateral and posteromedial thalamic nuclei combine to form the the lateral forebrain system, which is thought to represent the location, intensity, and quality of painful stimuli.
(3) The anterior cingulate cortex and the mediodorsal and parafascicular thalamic nuclei combine to form the medial system, which is thought to encode the affective dimension of the painful stimulus.
Using various pharmacological and genetic manipulations, the researchers inactivated various parts of these systems during the training phase and the recall phase (24 hours later).
First, they injected lidocaine in the anterior cingulate to temporarily make it non-functional. In a control group, they injected saline into the anterior cingulate, which should not have had any effect. Would chemically knocking out the anterior cingulate have an effect?
Similar effects were seen when lidocaine was administered to the mediodorsal or parafascicular thalamic nuclei (which feed into the anterior cingulate). These results indicate that the medial system is crucial for observational fear learning. However, when the lateral system was knocked out, there was no effect for observational fear learning. Therefore, fear learning requires the medial affective system to be intact, and not so for the lateral sensory system. Further similar experiments led to the conclusion that long-term memory storage of observational fear learning happens in the amygdala, and not the anterior cingulate. Further, knocking out the anterior cingulate had no effect on the ability of the demonstrator mouse to form the fear response when given electric shocks, only on the ability of the observer mouse to form an empathic fear response.
It gets even better!
They created genetic knock-out mice with non-functional anterior cingulate cortices. They messed with genes governing the calcium-ion channels, such that the neurons in the anterior cingulate were less excitable (so they didn't fire as much), and such that synaptic transmission between neurons in the anterior cingulate was disrupted. The knock-out mice had impaired empathic fear responses as well! Importantly, the knock-out mice still felt pain, and still formed a fear response when subjected to electric shocks - it was only the ability to form the empathic fear response which was disrupted. The knock-out mice also showed levels of anxiety and innate fear similar to normal mice.
So what does it all mean?
First, this is a really great study because they looked for converging evidence from behavioral, pharmacological, and genetic experiments. Really good science happening there.
Second, it confirms that the medial system (anterior cingulate and the thalamic nuclei that connect to it) is involved in the association of affective or emotional states with the fear response, not in producing the fear response itself based on sensory information.
Third, that the knock-out mice showed normal levels of anxiety and innate fear, but disrupted observational social fear suggests that there are different genetic and neural mechanisms that underlie these two different types of fear.
Fourth, that the fear response in the observer depended on the social relationship between the two mice suggests that the empathic fear response is not automatic or reflexive (like when babies automatically cry upon hearing other crying babies). There's a social-cognitive component!
Fifth, that there was a response (though a dulled one), even when there was an opaque partition between the two mice, suggests that other cues than visual (olfactory and auditory) should be taken into account. Even when the partition was opaque, the response was still stronger for siblings or mates than for strangers. Visual cues still seem strongest, however.
Sixth, the fact that these proto-empathic responses (even if not full-fledged empathy) are seen in mice suggests that empathy has some pretty evolutionarily old roots.
Finally, and this is getting speculative, some psychiatric disorders like psychopathy and sociopathy feature impaired recognition of the emotions or feelings of others. Abigail Marsh, who does research with psychopaths and other antisocial individuals, recounted a story about a colleague in an interview:
Marsh relayed a chilling anecdote about a colleague of hers, University College London psychologist Essi Viding, who was going through a task with a psychopathic murderer in which a series of faces with different emotional expressions were laid out before the woman. When the murderer saw the picture of the fearful face, she scratched her head and said: "I don't know what that expression is called, but I know it's what people look like right before I stab them."
Psychopathologists should investigate the role of the anterior cingulate system in these disorders, and the genes that govern the organization of that system.
Jeon D, Kim S, Chetana M, Jo D, Ruley HE, Lin SY, Rabah D, Kinet JP, & Shin HS (2010). Observational fear learning involves affective pain system and Ca(v)1.2 Ca(2+) channels in ACC. Nature neuroscience, 13 (4), 482-8. PMID: 20190743