In his essay on the writings of Leo Tolstoy, philosopher Isaiah Berlin [1] proposed two types of historical thinkers: those seeking to explain the world through unified principles (hedgehogs) and those seeking to approximate it through various, sometimes disparate, paths of understanding (foxes). As a metaphor for two different approaches to prediction, Berlin's idea raises an intriguing tension between explanation and approximation of natural phenomena. With a new decade of the brain dawning in psychiatry, a similar tension appears to underlie two different ways that neuroscientific thinking encroaches upon clinical understanding and practice: for simplicity, one can be termed a mechanismoriented approach and the other, an outcome-oriented approach. These approaches differ in terms of modeling principles, research methods, implications for drug discovery, and skill set needs (Table 1). The contrast between these two approaches raises interesting considerations for the field of psychiatry on scientific and clinical levels.
Mechanism-Oriented ApproachMechanisms in traditional biology are deterministic in nature: part A causes, permits, or inhibits part B in a biological process, and perturbations in a part will result in predictable changes to the biological state. These parts could be proteins in an intracellular signaling cascade, for example, or cortical nuclei in a neural circuit. Predictions are based on hypothesis testingexperiments where the mechanistic model is interrogated by functional changes to specific parts, as in a genetic mutation or drug inhibiting a protein's function, or the surgical ablation of a neuronal population. These experiments generate a unifying model of the biological process in question. The hope here is that therapies can be developed to target a specific biological component ("target-based" drug discovery) for a beneficial clinical effect. In certain subfields of oncology, a cancerous mutation is identified and a pharmacological agent is found that targets this oncogenic component in patients-as in the successful drug imatinib, which inhibits an overactive enzyme found in patients with chronic myelogenous leukemia [2].This hope is also reflected in recent priorities of the National Institute of Mental Health (NIMH) [3]. Neurobiological models of disease, validated by biological experiments to have high explanatory power, should guide the discovery of targeted therapeutic interventions for various psychiatric disorders. A clinical trial demonstrating the effectiveness of these interventions should, in principle, also show the expected mechanistic effects on neurobiological targets [3]. Pharmaceutical companies have extended this idea to "proof of concept" clinical trials, where new drugs developed through targetbased discovery are tested on small homogeneous populations of target-defined patients [4]. For example, a compound that targets a protein regulated by the Fragile X gene product can be tested in a subpopulation of patients carrying the same genetic and epigenetic characteri...