The early gene knockout studies with a neurobiological focus were directed at fairly obvious target genes and added very little to our knowledge of behavioural neuroscience. On the contrary, since the behavioural consequences were often predictable, this helped confirm that the technology was working. However, a substantial number of knockouts of genes expressed in the brain have been without obvious behavioural consequences, supporting the concept of genetic canalisation and redundancy. Others have produced a behavioural deficit for which there is no obvious explanation. Many cells of different tissue types have a capacity for memory, and in the brain, cells of the hippocampus are important for spatial learning and memory. Deleting genes that are expressed in the happpocampus has received considerable attention in this behavioural context. Although the initial studies experienced problems of interpretation, considerable advances have since been made. Knockout mice are now subject to tests of different forms of learning, multicellular hippocampal recordings, and restricted gene deletion specific to cells of component regions. This multi‐level approach is proving more informative. Nevertheless, there is still a need to recognise that behavioural expression is several steps removed from gene expression, and that the relationship between genes and behaviour can be reciprocal.