While the important role of psychosocial factors in the development of criminal behavior has long been acknowledged, there has been an increasing interest in the neurobiological basis of aggression and crime over the past decade, boosted by methodological advances in neuroimaging and molecular genetics (1, 2). Researchers recognized the potential role of biological factors in the etiology of criminal behavior some time ago, and in this vein an intriguing result of a long-term study is reported in this issue of the Journal. Gao and colleagues (3), using data rooted in a study that was initiated more than 20 years ago, tested whether abnormal fear conditioning predisposes to crime. Fear conditioning is a basic form of learning in which fear is associated with a previously neutral stimulus. Experimentally, this is done by pairing a neutral stimulus with an aversive one, such as an unpleasant loud noise. Eventually the previously neutral stimulus alone will evoke a fear response that can be quantified with physiological measurements such as electrodermal responses.

The study by Gao et al. is based on the hypothesis that fear conditioning is the mechanism by which we learn to link antisocial acts with negative consequences such as punishment or social exclusion. By leading to a failure in such social learning, poor fear conditioning would thus predispose an individual to antisocial behavior. Crucially, if poor fear conditioning has a causal role in crime, it should be detectable early in life, before antisocial behavior becomes manifest. In the framework of a large birth cohort study, Gao et al. tested fear conditioning in children at age 3. Twenty years later, they probed the association of poor fear conditioning in early childhood with adult criminal behavior. A group of 137 individuals who had been convicted of committing crimes by age 23 was compared with a carefully matched group from the same cohort who showed no evidence of criminal behavior. Strikingly, skin responses to the conditioned stimulus were significantly smaller in children who became criminal later on. In fact, on average there was no evidence for any conditioning at all in the criminal group, as skin responses to the conditioned stimulus were statistically indistinguishable from responses to a stimulus that was not conditioned. The implications of this finding are at least twofold. First, it provides new insights into the etiology of criminal behavior; second, it may help to guide early therapeutic interventions in children at risk for antisocial behavior.

The central brain structure in the circuitry underlying fear conditioning and the perception of threatening stimuli in the environment is the amygdala (4). Deficient amygdala function has been proposed to render individuals unable to recognize cues that signal threat, making them relatively fearless (5). Less sensitive to any negative consequences of their behavior, fearless individuals engage more readily in antisocial behavior. Indeed, the relationship between antisocial behavior and amygdala dysfunction is supported by a number of findings in adults and adolescents, including deficits in recognizing fearful facial expressions, poor fear conditioning, and reduced amygdala responses to negative emotional stimuli as measured with functional neuroimaging (2, 5, 6). While such findings in adults and adolescents are ambiguous as to whether they represent cause or consequence of repetitive antisocial behavior, the results reported here by Gao et al. now point to an early deficit in amygdala function as a causal mechanism. However, the notion of a generally hyporesponsive amygdala would be too simplistic. The amygdala is a small yet complex structure comprising a number of subnuclei with distinct functional properties (4). Moreover, some studies have shown increased amygdala responsiveness in antisocial individuals (2) and in those with a genetic predisposition to aggressive behavior (7). Contrasting with the fearlessness hypothesis outlined above, such findings point to a different pathomechanism whereby pathological aggression may be related to heightened anxiety. Increased amygdala responsiveness is a well-established finding in anxiety disorders and is thought to be the basis for exaggerated fear conditioning. This is nicely demonstrated by another study published in this issue. Using a simple but elegant modification of classical fear conditioning, Lissek et al. (8) show that in panic disorder fear responses are more easily generalized—that is, fear responses are evoked more easily by stimuli that resemble the conditioned stimuli in patients with panic disorder compared with healthy individuals.

The heterogeneity of findings from neuroimaging studies suggests that the amygdala might be differentially involved in different etiologies of antisocial behavior. This exemplifies the necessity of taking into account the whole spectrum of possible pathogenetic factors and developmental pathways if we are to develop a more complete understanding of antisocial behavior. Especially when it comes to developing targeted preventive or therapeutic interventions, it will be crucial to identify subgroups of at-risk individuals in whom different neurobiological mechanisms underlie antisocial behavior (5, 9). The Gao et al. study (3) is an important step toward this goal, as it identifies a biological marker that may help to identify subgroups with a distinct neurobiological profile even in early childhood. Clearly, the scope for changing behavior will be greatest in the early years because of the greater plasticity of the brain in childhood.

Is poor fear conditioning at age 3 likely to represent an innate trait? While this question cannot be answered with certainty, it is clear that a wealth of environmental influences can already have taken effect during the first 3 years and even in utero. Reduced fear conditioning in 3-year-olds may thus represent not just a genetic predisposition but rather the early manifestation of gene-environment interactions (9). The authors of the study aimed to minimize the chances that differences in fear conditioning were due to adverse environmental factors by matching the two groups for social adversity. However, some important issues were not systematically assessed and may have gone unnoticed, such as differences in nutrition, parenting style, and maternal stress during infancy. The latter is indeed associated with behavioral problems later in childhood, as shown in another carefully conducted longitudinal study published in this issue (10).

There are a number of caveats when it comes to drawing conclusions from and considering the potential consequences of findings such as those reported by Gao et al. First, there is a tendency, not only among scientific laypersons, to mistake neurobiological findings for evidence of a fateful and unchangeable condition. It is of the utmost importance that this view be acknowledged as wrong. One of the most stupendous characteristics of the nervous system is its plasticity, and neurobiological measures thus often represent little more than a snapshot at one stage in development. Just as longitudinal assessment of behavior elucidates developmental pathways to psychological problems (10), longitudinal studies of neurobiological factors could help us better understand their contribution to long-term behavioral outcome. Second, while poor fear conditioning can be seen as a predisposing factor for criminal behavior, it can by no means be used to predict future behavior. Such markers can only be applied at the group level and are far from being selective or specific enough to be used as diagnostic or screening tools in individuals. Finally, crime is a multifaceted behavioral outcome of complex interactions among multiple biological and environmental factors and cannot possibly be explained by a single neurobiological factor such as fear conditioning.

If not handled with great caution, neurobiological markers can easily be misused to stigmatize individuals who are perceived as a potential threat to society. With the increasing availability of data that help us prevent, diagnose, and treat antisocial behavior early in life, we also need a public debate on how to use this information and, even more important, how to avoid its misuse. Neurobiological research offers a great chance to further our understanding of antisocial and criminal behavior. This understanding should be used to benefit those children who are at greatest risk for a criminal career and to design interventions that are tailored to their needs.