38 Motivation
by the obesity of rats given a variety of highly palatable foods
(Sclafani & Springer, 1976). Rats under these conditions can
more than double their weight and behave similarly to ani-
mals that have obesity-inducing brain lesions.
These findings do not undermine the alliesthesia model of
food selection. Rather, they suggest that exposure to the sen-
sory aspects of food, in the absence of ingestion, is sufficient
to reduce the palatability, and therefore intake, of that food. A
variety of studies demonstrated just such a result. Changes in
the shape of the food have an effect on intake. Rolls, Rowe,
and Rolls (1982) showed that subjects would consume more
pasta if it were offered as spaghetti, half hoops, and bow ties
than if it were offered as spaghetti alone. Guinard and Brun
(1998) demonstrated that variation in another nonnutritive di-
mension, food texture, can similarly lead to increases in con-
sumption. Rolls and Rolls (1997) have demonstrated that
chewing or smelling food is sufficient to induce alliesthesia-
like reductions in the subsequent palatability of that food
in the absence of eating that food. Thus, although ingestion
may be sufficient to cause alliesthesia, it is not necessary:
Sensory stimulation alone is sufficient to cause changes in
palatability and to induce variety in food choice.
Factors Governing the Incentive Aspects of Foods
Cathexes
The regulation of feeding behavior through meal patterning
and the regulation of food variety through alliesthesia assume
that the animal knows which stimuli present in the environ-
ment are foods that will satisfy its nutritional requirements.
In the case of opportunistic omnivores such as humans and
rats, this knowledge must be learned. This process was de-
scribed as the development of cathexes by Tolman (1949),
who suggested that it involved affective, or emotional, learn-
ing that created positive affective reactions toward sub-
stances that fulfilled nutritional needs and negative affective
reactions toward substances that did not or that caused un-
pleasant reactions such as nausea. Learning of negative
cathexes has been the more fully explored of these processes
through examination of conditioned taste (or flavor) aversion
(CTA).
Exploration of CTA has demonstrated a distinction be-
tween aversive motivation caused by insults to the skin
defense system, such as electric shock, and insults to the gut
defense system caused by taste and emetic toxins (Garcia y
Robertson & Garcia, 1985). This suggests that learning about
the incentive value of food is based on selective associations
between taste (and to a lesser extent olfactory stimuli) and
postingestive consequences. However, in many cases the
animal must make behavioral choices at a distance, before
being in a position to taste the potentially aversive food. A
great deal of research suggests that associations between the
distal cues that guide behavior and the postingestive conse-
quences of ingesting a food predicted by those cues require
mediation by taste or olfactory cues (Garcia, 1989). This sug-
gestion gives rise to a mediated-association view of food in-
centive learning: Postingestive consequences are associated
with taste, and taste stimuli are associated with distal cues.
Hence, feeding behavior is governed by a chain of distal
cue–taste–postingestive consequence associations (Garcia,
1989).
The strongest evidence for this view comes from a variety
of studies that emphasize the importance of taste in mediating
CTA to distal cues. Rusiniak, Hankins, Garcia, and Brett
(1979) demonstrated that although weak odor paired with
nausea produces weak aversion to the odor, the same odor re-
sults in a much stronger aversion if presented in compound
with a taste. Brett, Hankins, and Garcia (1976) demonstrated
that after repeated trials, hawks rejected both black (poi-
soned) and white (safe) mice, but that following the addition
of a distinctive taste to the black mice, the hawks began to re-
ject the black mice and eat the white mice. Evidence also
suggests that similar, though weaker, effects can be found by
using the expectancy of a taste to mediate the CTA to distal
cues. Holland (1981) paired a tone (distal) CS with a distinc-
tive flavor before pairing the tone with a nausea-inducing
lithium chloride injection. Subsequent testing showed de-
creased consumption of the tone-predicted food, indicating
the development of an indirect, expectancy-based CTA.
Taken together, these results indicate that learning about
which foods in the environment to ingest is mediated by two
different Pavlovian conditioning processes.Incentive LearningAlthough this system indicates to the animal in a general
sense what is good to eat, it is not able to guide the animal’s
day-to-day foraging behavior because the gustatory learning
system proposed to underlie cathexes is purely affective; it
encodes only positive or negative values. To the extent that
an animal’s behavior reflects its current needs, the animal
must be able to encode and act on the value of food given its
current internal state. The evaluation of the incentive value of
food given the animal’s current internal state is called incen-
tive learning(Balleine, 1992).
The study of incentive learning is complicated by the
fact that the effect of internal state on feeding responses
seems to differ based on the associative procedure that is
used to examine those behaviors. In Pavlovian conditioning