Several lines of evidence suggest that this early olfactory conditioning may be a special form of learning, unique to the immature olfactory system. For example, early appetitive olfactory learning in rats is acquired very rapidly (Sullivan & Leon, 1987) during a sensitive period after birth (Woo & Leon, 1988) and lasts into adulthood (Coopersmith & Leon, 1986; Fillion & Blass, 1986). Thus, early olfactory learning in rat pups could be similar to imprinting to visual stimuli in birds (Horn, 1985) or imprinting to chemical stimuli in fish (Scholz, Horral, Cooper, & Hasler, 1976).

Further evidence in support of unique mechanisms of early olfactory learning arise from the nature of the neural changes associated with it. Olfactory bulb responses to appetitively conditioned odors are characterized by enhanced focal metabolic activity in odor-specific regions of the glomerular layer (2-deoxyglucose [2-DG] uptake) and modified response patterns of olfactory bulb output neurons (mitral-tufted cells) near these hyperactive glomeruli (Sullivan & Leon, 1986; Wilson & Leon, 1988). In addition, the size of the glomerular layer near the regions of hyperactivity is increased after early odor experience, suggesting either an experience-associated increase in neural size or number or decrease in normal postnatal death of these neurons (Woo, Sullivan, & Leon, 1987).

Associative olfactory conditioning consisted of seventeen 10-min training sessions with an intertrial interval (ITI) of 24 hr. On PN 1, pups were assigned to one of the following training groups: (a) forward odor-stroke, pups were exposed to peppermint odor while being vigorously stroked with a sable hair brush (13 normal and 10 extinction for behavior; 7 normal and 7 extinction for 2-DG; 4 normal and 3 extinction for physiology); (b) random odor-stroke, pups were given nonoverlapping exposures to both odor and stroking using a pseudorandom schedule (13 normal and 10 extinction for behavior; 7 normal and 7 extinction for 2-DG; 5 normal and 3 extinction for physiology); (c) forward odor-shock, pups were exposed to the peppermint odor paired with a footshock coinciding with the last second of odor presentation (13 normal and 13 extinction for behavior; 7 normal and 7 extinction for 2-DG; 4 normal and 3 extinction for physiology); (d) random odor-shock, pups were given nonoverlapping exposures to both odor and shock using a pseudorandom schedule (13 normal and 13 extinction for behavior; 7 normal and 7 extinction for 2-DG; 4 normal and 3 extinction for physiology); or (e) naive (15 for behavior; 7 for 2-DG; 4 for physiology). The tactile stimulation produced by stroking was used to mimic maternal stimulation and has reinforcing properties as robust as those of milk to infant rats (Sullivan & Hall, 1988). The intensity of the stroking was sufficient to induce behavioral activation in the pups. The footshock was a 1-s, constant-current, 1.5-mA shock. This shock intensity has been shown to be capable of producing an odor aversion throughout development (Kucharski & Spear, 1984; Sullivan, 1990; Sullivan & Wilson, 1990). The odor for all shock and stroking groups was peppermint extract (Shilling, Baltimore, MD; the alcohol in this peppermint oil and alcohol mixture was allowed to evaporate before use) presented through a flow-dilution olfactometer (concentration 1:10 of saturated vapor; 1 L/min flow rate). On PN 18 and PN 19, half of the pups from each group were given six extinction trials: 10-min odor exposures, 2-hr ITIs, 3 times per day.

On PN 20, pups were given one of three tests: a behavioral odor preference test, measurement of ^l4C-2-DG uptake during conditioned stimulus-only exposure, or measurement of mitral-tufted cell single unit response patterns to the conditioned stimulus.

The behavior test consisted of a Plexiglas Y maze (start box: 8.5 cm width, 10 cm length, 8 cm height; choice arms: 8.5 × 24 × 8 cm) that required pups to choose between two odors:—the conditioned peppermint odor and a familiar pine odor. Both odors were delivered via flow-dilution olfactometers (1:10 concentration of saturated vapor; 1 L/min flow rate) into the ends of the choice arms. Pups were placed in the start box and allowed 60 s to choose an arm. The pups were removed from the maze for 30 s between trials and received five trials. Odors were exhausted from the maze between trials.

For the odor-2-DG test, pups were injected with ¹⁴C-2-DG (20 μCi/100 g) immediately before odor delivery and placed in a test chamber identical to that used during conditioning. Peppermint odor was delivered at 1 L/min (1:10 dilution) for 45 min (cycle of 3 min on, 1 min off). After odor exposure, pups were decapitated, and their brains were quickly removed and frozen in methylbutane at -45 °C. The frozen brain was equilibrated to -17 °C in a cryostat for 45 min, and the olfactory bulb was cut coronally in 20-μm sections. Brain sections were exposed on Kodak SB-5 film for 8 days at room temperature, and autoradiographs were analyzed with an image analysis system (Imaging Research, Inc., St. Catharines, Ontario, Canada) as previously described (Sullivan, Wilson, & Leon, 1989; Wilson & Sullivan, 1990).

Despite the opposing conditioned behavioral responses in forward odor-stroke and forward odor-shock pups, both conditioning procedures produced similar modifications in olfactory bulb metabolic and neural activity (Figures (Figures22 and and3).3). Forward odor-stroke and forward odor-shock pups demonstrated enhanced focal glomerular layer 2-DG uptake compared with random and naive controls. Relative focal 2-DG uptake in odor-specific regions of the glomerular layer was significantly enhanced in both forward odor-stroke and forward odor-shock pups, 2 × 5 ANOVA: Training × Extinction interaction, F(4, 60) = 6.9, p < .001. Post hoc Tukey tests revealed that focal 2-DG uptake in forward odor-shock and forward odor-stroke pups was significantly different from controls, p < .01. Relative 2-DG uptake in nonfocal glomerular layer areas did not significantly differ between groups. Extinction training eliminated the enhanced focal 2-DG uptake patterns in forward odor-stroke and forward odor-shock pups. Post hoc tests revealed that 2-DG uptake in forward odor-shock extinction and forward odor-stroke extinction pups were significantly different from nonextinction pups and not significantly different from controls, p < .01.

Of importance, both the behavioral and neural responses to conditioned odors could be extinguished. This suggests that early olfactory learning is not a special form of learning or imprinting but rather conforms to criteria of normal associative conditioning. The reversal of conditioned neural changes by extinction found in the present study is similar to results obtained in mature cats and rabbits (Applegate, Frysinger, Kapp, & Gallagher, 1982; Diamond & Weinberger, 1989; Woody & Brozek, 1969). It should be noted, however, that given the techniques used here, it is not clear whether the similarity between the extinguished responses and responses in control animals are due to similar mechanisms; that is, we do not know whether extinction training eliminated the modifications produced by conditioning or whether extinction produced additional changes that resulted in responses being similar to those of control animals.

The present results suggest that although olfactory bulb metabolic and neural responses to odors are modified by conditioning, the modified response does not distinguish between learned odors signaling approach and those signaling avoidance. This result was found despite dramatic differences in learned behavioral responses to that odor. Although we are unaware of a direct comparison, these results appear to compare favorably with those obtained in mature rabbits. In two separate studies using adult rabbits, olfactory bulb electroencephalographic burst amplitude was reduced to odors previously conditioned to either aversive shock (Bressler, 1988) or water reward (Viana Di Frisco & Freeman, 1985).

The mechanisms of change in olfactory bulb response are, as yet, unknown. A variety of previous studies have demonstrated that the enhanced 2-DG uptake to learned odors in pups is not due to modified respiration (Coopersmith & Leon, 1984; Sullivan, Wilson, Kim, & Leon, 1988). Furthermore, the changes in both focal 2-DG uptake and single unit activity have been shown to be specific to the conditioned odor; neurobehavioral responses to nonconditioned stimuli are not modified by associative conditioning in pups (Coopersmith, Henderson, & Leon, 1986; Wilson ctal., 1987). However, it has not been definitively demonstrated whether the change in bulb response is due to synaptic changes intrinsic to the bulb or is due to centrifugal modulation of bulb activity. Although the present article does not directly address this issue, there are several lines of evidence suggesting that neural changes intrinsic to the bulb do occur after early olfactory experience. First, odor-stroke pairings in pups induce an increase in glomerular size near odor-specific 2-DG foci (Woo et al., 1987). Second, blockade of olfactory bulb/β-noradrenergic receptors with intrabulbar infusions during training blocks acquisition of learned odor preferences (Lin, Wilson, & Sullivan, 1990). Third, although noradrenergic centrifugal projections to the bulb appear to be required for acquisition of learned olfactory behaviors in pups (Sullivan et al., 1989), norepincphrine is not necessary for expression of the learned response (Lin et al., 1990). However, given the inability of the bulb to distinguish between learned cues signaling opposite behavioral responses, as shown here, it is clear that other brain regions must be modified or at least participate in the learned response.

Based on the present results, we hypothesize that the acquisition of even a simple association during olfactory conditioning in rat pups requires the subject to store at least two bits of information. One bit signals that a stimulus has acquired significance (“I know that odor”), and the second bit signals what the significance is (“approach-avoid”). The results of the present experiment suggest that the first bit may be encoded by activity in the olfactory bulb—the bulb responds differentially to learned odors. Other brain regions (Kucharski & Hall, 1987, 1988), however, must be involved in the encoding of the hypothesized second bit, given that bulb activity cannot distinguish learned preferred odors from learned aversivc odors. One possible locus for information regarding this remaining information is the amygdala. The amygdala receives direct olfactory input (Price, 1987), and amygdala neurons respond to olfactory stimuli (Cain & Bindra, 1972). Furthermore, the amygdala has long been associated with learning and memory, and damage to this area can impair associative learning (Sarter & Markowitsch, 1985), even in infant rats (Sananes & Campbell, 1989). The olfactory cortex, anterior olfactory nucleus, frontal cortex, hippocampus, and a variety of other areas have also been implicated in olfactory memories.