_There are two main reasons, the one internal and the other external, why the reaction experiment has not developed along the lines of our psychological classification of action._ The internal reason is that _the reactor is extremely sensitive to slight changes in instruction_, in the rules laid down for the experiment. We have already had an instance: the sensory reaction is a skeleton impulsive action; but the motor reaction, which results from a shift of emphasis in the instruction, is not sensorimotor; it is an abbreviated or telescoped impulsive action. Psychologists have naturally been interested in this side of the experiment, and so have tried the effect of varying instructions, instead of duplicating in the laboratory the gross types of action that our classification distinguishes. The second, external reason is that _the reaction, largely on account of its outside origin, was for some time treated in a chapter apart_; not until the nineties of the last century did psychologists realise that it gave them experimental control of action; and so the technique has been complicated and the outline filled in without special reference to the psychology of action. We need not here go into details; it is enough to say that experimenters have tried the effect of increasing the number of stimuli, and thus of leaving the reactor more or less uncertain of what he shall expect; of increasing the number of possible responsive movements; and of varying the instruction given beforehand to the reactor, in such wise that a particular responsive movement is assigned to a particular stimulus, or that response is made to certain stimuli but not to certain others. _All these forms of compound reaction have an interest of their own_, which makes their analysis desirable; they enable us to trace the establishment and course of determining tendencies, the tendencies set up by the instructions; and some of them throw light upon the psychology of negative instruction (p. 250). Only, as we have said, they do not represent the different types of action. Things are now changing; but a great deal of work must be done before we obtain typical analyses of the actions discussed in the preceding paragraphs.