As an undergraduate at Northwestern University, Hovland
acquired a strong background in mathematics, physics, and
biology, as well as in experimental psychology, receiving
his A.B. with highest distinction in 1932 (just before turn-
ing twenty). On a Catharine White fellowship he also ob-
tained his A.M. there in 1933 and completed research that
appeared in his earliest published papers (the first, coau-
thored with a stimulating new Northwestern faculty mem-
ber G. L. Freeman on "diurnal variations in performance
and related physiological processes").
Concerning a letter recommending Hovland for gradu-
ate study, Yale's Walter R. Miles recalled that, "The letter's
language of so high approval and praise was such as to
make [the] Yale professors smile and shake their heads. As
events evolved they were using similar language in . . .
recommending the same Carl Hovland . . . a very few years
later" (Miles, 1961, p. 122). Hovland prepared six papers
for publication during his first year and in just two more
years he received his 1936 Ph.D. with honors under the
prominent Yale learning theorist Clark L. Hull.
Hovland's dissertation provided the first evidence for a
law of generalization, in which the tendency to make a
response learned to one stimulus falls off exponentially
with the distance separating a test stimulus from the origi-
nal training stimulus along a sensory continuum, such as
the continuum of auditory pitch (Hovland, 1937). Begin-
ning with my own dissertation twenty years later, I devel-
oped a new approach that provided more definitive evi-
dence for such a law (Shepard, 1958, Figure 2) and, thirty
years still later, a theoretical justification for the law's pos-
sible "universal" character (Shepard, 1987, Figures 1 and
3). Such a law of generalization was also central to the
interpretation of the results of our joint study of classifica-
tion learning (Shepard, Hovland, and Jenkins, 1961, pp.
25-30). I still regard generalization as the most fundamen-
tal problem of human, animal, and machine learning-if
not, indeed, of education and cognitive science generally.
On completing his dissertation, Hovland was immedi-