Carlsson has put forward the hypothesis that the positive and negative symptoms of schizophrenia are due to failure of mesolimbic and mesocortical projections consequent on hypofunction of the glutamate N-methyl-d-aspartate (NMDA) receptor. The hypothesis has been recently emphasized in this Journal that the loss of synaptic spines with NMDA receptors, which can be precipitated by stress, can explain the emergence of positive symptoms such as hallucinations and that this synapse regression involves molecules such as neuregulin and its receptor ErbB4 that have been implicated in schizophrenia. In this essay these two hypotheses are brought together in a single scheme in which emphasis is placed on the molecular pathways from neuregulin/ErbB4, to modulation of the NMDA receptors, subsequent changes in the synaptic spine's cytoskeletal apparatus and so regression of the spines. It is suggested that identification of the molecular constituents of this pathway will allow synthesis of suitable substances for removing the hypofunction of NMDA receptors and so the phenotypic consequences that flow from this hypofunction.