To achieve the improvements advocated by Emily Holmes and colleagues for psychological treatments (Nature 511, 287–289; 2014), researchers need to conceptually link studies of specific psychiatric disorders with fundamental processes that are shared by different disorders.

Psychologists often manipulate the environment of study participants (the independent variable) to alter a person's response or behaviour (the dependent variable). For example, they might compare the effects of threatening or neutral images on a subject's physiological arousal or memory. This approach lends itself to statistical analysis of group data, but it overlooks the important point that humans already control their environment by altering their responses. An example would be an anxious person who actively seeks safety by avoiding eye contact.

Research methodologies need to take into account the fact that such negative-feedback mechanisms exert control at all levels, including physiological, psychological and social (see T. A. Carey Lancet 382, 1403–1404; 2013).