Feeding 37of small meals and the frequency of short intermeal intervals
decreased as the response requirement increased, leading to a
smaller number of larger meals and the conservation of total
intake and body weight.
Similar effects of meal-procurement cost have been
demonstrated across a variety of animal species with a vari-
ety of evolutionary niches and foraging strategies (Collier &
Johnson, 1990). The determination of meal cost appears to be
calculated by the animal across a relatively long time win-
dow: Animals trained on alternating days of high and low
cost learned to feed primarily on low-cost days (Morato,
Johnson, & Collier, 1995). Animals also show a nonexclusive
preference for feeding on low cost resources (Collier, 1982),
on larger pellets where the cost is the same as for smaller pel-
lets (Johnson & Collier, 1989), and on pellets with higher
caloric density (Collier, Johnson, Borin, & Mathis, 1994).
Animals also include risk of aversive events into the cost
equation. Fanselow, Lester, and Helmstetter (1988) demon-
strated that increased numbers of randomly occurring foot
shocks led to changes in meal patterning similar to those
induced by increased procurement costs. Characteristics of
feeding demonstrated in session and free-feeding procedures,
such as increased rates of responding or consumption or cor-
relations between length of food deprivation and subsequent
meal size, are not replicated in the laboratory feeding proce-
dure (Collier & Johnson, 1997; Collier et al., 1999). This se-
ries of results has led Collier and his coworkers to suggest
that the crucial determinants of feeding initiation are the costs
associated with meal procurement and that physiological
functions act to buffer the effects of variations in feeding ini-
tiation determined by procurement cost rather than as the in-
stigators of feeding behavior (Collier, 1986).
The Behavioral Ecology of Feeding Cost
In the laboratory, costs are determined by the experimenter. In
the real world these costs are determined by the animal’s eco-
logical niche, placing feeding behavior under the direct con-
trol of evolutionary factors. Feeding intensity can be predicted
from relative predatory risk, as can be inferred from the study
by Fanselow et al. (1988). For example, large predators could
be expected to eat long-duration, low-intensity meals because
they are not subject to threat from other animals. In contrast,
small predators could be expected to eat short-duration, high-
intensity meals as they are themselves potential prey. These
suggestions are consistent with ethological data (Estes,
1967a, 1967b; Schaller, 1966). Meal patterning and feeding
initiation can be predicted from food type. Predators could
be expected to sustain high procurement costs for their nutri-
tionally rich meals, whereas herbivores—particularly small,
monogastric herbivores—could be expected to take frequent
meals because of the low quality and intensive processing
required by their usual foods. These suggestions have been
supported by experimental data indicating that cats can eat
every three to four days when procurement costs are high
and maintain bodyweight, whereas guinea pigs are unable to
maintain their bodyweight with fewer than two to three meals
per day and are unable to sustain high procurement costs
(Hirsch & Collier, 1974; Kaufmann, Collier, Hill, & Collins,
1980).Factors Governing Variety of IntakeAlliesthesiaFood selection must provide all the nutrients necessary for
survival. This task is simple for a specialized feeder that eats
very few foods. However, opportunistic omnivores such as
rats and humans contend with a potentially bewildering array
of choices. Traditional approaches have suggested that the
body detects hunger when it is deprived of a particular com-
modity, and this homeostatic need sets in motion behaviors
directed at correcting the deficit (e.g., Rodgers, 1967). Thus,
intake of various nutrients could be regulated by set points
for these nutrients. Food palatability had been suggested to
be an alternative mechanism (Mook, 1987). Assume that an
animal (or at least an opportunistic omnivore) eats because
food tastes good. If that is combined with one other assump-
tion, that food loses its incentive value when consumed, we
have a mechanism that ensures intake of a variety of sub-
stances. This phenomenon is referred to as alliesthesia
(Cabanac, 1971). Cabanac demonstrated that palatability
ratings of sugar solution change from positive to negative
following ingestion, but not simply the taste of, sucrose.Sensory SatietyDespite this evidence, it is also true that sensory, rather than
postingestive, stimuli associated with food play an important
role in inducing variety of intake. The clearest demonstra-
tions of these effects are those demonstrating the effects of
food variety in sated animals and people. When we sit down
to our holiday meal, the turkey tastes exquisite, but after two
or three helpings we can barely tolerate another bite. Yet de-
spite our satiety, we proceed to eat a large dessert. The order
of courses does not matter (Rolls, Laster, & Summerfelt,
1991); the critical determinant of renewed consumption
is that the food has variety (Rolls, 1979). This variety effect
has been demonstrated in humans and rats (see Raynor &
Epstein, 2001, for a recent review), perhaps most dramatically