At present it is not clear how instrumental actions arise. In our experiments the mollusk Helix received an aversive stimulus when one of its neurons did not generate an action potential in response to a conditioned stimulus. The appearance of an aversive stimulus did not depend on the generation or failure of a spike in the control neuron. Although the trained and the control neurons were not coupled and did not connect directly by synapses, the magnitudes of the conditioned responses of both the trained and the control neurons depended on the relationship between the nearest previous neuronal responses and the rapidity of their change. These relationships changed several times during learning, and depended on the initial excitability of the control neurons. The trial-and-error method may have revealed itself at the neuronal level.