Table 2.

Summary of results for studies examining goal-directed learning

	Unmedicated a	Goal-directed control	Sample size (case, control)	Task
Obsessive compulsive disorder
Gillan et al. (2011)		↓	20, 20	Devaluation
Gillan et al. (2014b)		↓	25, 25	Devaluation
Gillan et al. (2014a)		↓	20, 20	A-O learning
Voon et al. (2014) b	×	↓	32, 96	MB learning
Gillan et al. (2015a) c		↓	37, 33	Devaluation
Social anxiety disorder
Alvares et al. (2014)		↓	23, 23	Devaluation
Alvares et al. (2016) b		↓	20, 19	Devaluation
Schizophrenia
Morris et al. (2015)		↓	18, 18	Devaluation
Addiction
Sjoerds et al. (2013) [alcohol]		↓	31, 19	A-O learning
Voon et al. (2014) [meth] b		↓	22, 66	MB learning
Voon et al. (2014) [alcohol] b	×	n.s.	30, 90	MB learning
Ersche et al. (2016) [cocaine]		↓	72, 53	Devaluation
Tourette syndrome
Delorme et al. (2016)	×	↓	17, 17	Devaluation
Delorme et al. (2016)		n.s.	17, 17
Autism spectrum disorders
Geurts & de Wit (2013) d		n.s.	24, 24	Devaluation
Alvares et al. (2016) b		↓	17, 19	Devaluation
Eating disorders
Voon et al. (2014) [binge eating] b	×	↓	31, 93	MB learning

n.s., Non-significant.

Goal-directed learning was measured using devaluation, model-based (MB) learning or action-outcome (A-O) learning test. The latter two measures are proxies for devaluation sensitivity (Gillan et al. 2011, 2015b).

^aUnmedicated were free of selective serotonin re-uptake inhibitors or antipsychotics for at least 2 weeks, and free of stimulants for 17 h. In the case of addiction, unmedicated applies when subjects meet the above criteria and are abstinent.

^bCompared multiple diagnostic groups.

^cCompared medicated and unmedicated groups, and no medication effect was observed.

^dPaediatric sample.