A music learning experiment was performed, illustrating the possibility of isolating certain components of performance (labeled Skill and Motivation) by using separate measurement situations for each. Straightforward application of Hullisn theory-which considers that component factors of performance interact prior ro overt behavior and are not thus separable in measurement situations-to the experiments data led to a self-contradiction. It was shown, further, that Hull's theory could be considered consistent with the present data only when certain assumptions of dubious plausibility were made.The concept of performance is central to learning theory in that the learning process cannot be observed more directly ( a t present) than through the medium of the learner's performance, his overt behavior. In the context of a new approach to learning theory (Greenwald, 1964) it has been brought O L I~ that learning theorists have uniformly referred to variations in performance by reference to a "response-strength" type of construct, such as Hull's (1952) reaction potential or Rotter's ( 1 9 5 4 ) behavior potential. Such construccs used to describe performance have in common a single underlying dimension that ranges from "weak" to "strong." In other words, response tendencies are generally considered to be in a state of relative weakness or strength and this state is supposed to correlate monotonically with such observable response characteristics as latency, rate, amplitude, probability, or resistance to extinction. The new approach proposes that performance is best conceived of as having two conceptually separable and independently measurable dimensions or components, rather than the single "strength" dimension. These two proposed dimensions of performance and their measures may be briefly summarized as follows."