Sheep use preingestive cues as indicators of postingestive consequences to improve food learning

Previous work has shown that herbivoressuccessfully learn to associate food sensory characteristicswith postingestive consequences when theconditioning procedure is simple, whereas this abilitybreaks down when the learning task is made morecomplex. We hypothesized that sensory characteristicscould act as indicators of postingestive consequencesand that the presence of preingestive cues would improvethe food learning of sheep in situations varyingin complexity. Sixteen sheep were subjected to afirst conditioning phase to associate 2 flavors addedto alfalfa hay with either a positive or a negative consequence,induced by intraruminal administration ofstarch (330 mg/g of DMI) or LiCl (5 mg/g of DMI).Sheep progressively decreased their choice of the flavoredhay associated with the negative consequence (P< 0.05). This procedure provided sheep with experiencewith postingestive consequences associated withthe different flavors. In a second conditioning phase,the experienced sheep and 16 naïve sheep were dividedinto groups of 8 and subjected to either a simple or acomplex conditioning procedure [i.e., the 2 flavors wereoffered on separate days (simple conditioning) or simultaneouslywithin a day (complex conditioning)]. The 2flavors applied to grass hay were associated with eitherpositive (starch, 330 mg/g of DMI) or negative (LiCl,10 mg/g of DMI) consequences. As hypothesized, sheepin the simple conditioning group expressed a greateraversion to the flavored hay associated with the negativeconsequence than did those in the complex conditioninggroup (0.303 ± 0.035 vs. 0.474 ± 0.035 respectively;P < 0.01). Experienced sheep rejected theflavor associated with the negative consequence morestrongly than did naïve sheep, regardless of the conditioningprocedure (0.304 ± 0.029 vs. 0.470 ± 0.041respectively; P < 0.05). The initial increased preferencefor aniseed (0.80 ± 0.04), however, greatly influencedfood learning because sheep negatively conditioned onthis flavor expressed less avoidance than those negativelyconditioned on orange (0.53 ± 0.04 vs. 0.25 ±0.03 respectively; P < 0.01). In conclusion, the simultaneousscenario was actually perceived as complex byall sheep, and experienced sheep were more efficientin food learning than naïve sheep in both the simpleand complex learning contexts. The sheep were thusable to generalize the association between sensory cuesand postingestive consequences, especially in a complexenvironment, and then to use these sensory cues as indicatorsof postingestive consequences