Humidity as an aversive stimulus in learning in <Emphasis Type="Italic">Drosophila melanogaster </Emphasis>

Learning can be defined as a behavioral change linked to previous individual experience. Animal models are useful in the study of learning, and the fruit fly Drosophila melanogaster has been used for 30 years to this end (see Le Bourg & Buecher, 2002, for a review). Establishing paradigms for the study of learning is not an easy task and requires, in addition to replicability, accurate control and characterization of the stimuli driving the animal’s response. The present article is an attempt to offer a suitable characterization of the stimuli involved in a recently described paradigm (Le Bourg & Buecher, 2002). This paradigm consists in using a T-maze with one darkened and one illuminated arm to train individual D. melanogaster flies to suppress their natural positive phototactic tendency and choose the darkened arm preferentially. The lighted arm leads to a lighted vial containing a filter paper wetted with an aversive quinine solution, whereas the darkened arm leads to a darkened vial with no aversive stimulus. Over a 16-trial training session, young flies of both sexes increase their tendency to choose the darkened vial when the lighted vial is wetted with quinine. By contrast, most flies tested with a dry lighted vial do not increase their tendency to avoid this vial (Le Bourg & Buecher, 2002). The procedure has several advantages. First, it does not require any preparation of flies before training, such as fasting (as is the case with conditioned inhibition of the proboscis-extension response (PER; see, e.g., Brigui, Le Bourg, & Médioni, 1990). Second, it yields results quickly: A 16-trial training session takes about 15 min for each young fly. Third, the procedure does not require sophisticated devices that may be unavailable to many researchers (e.g., the flight simulator used to study visual learning in D. melanogaster; see Wolf & Heisenberg, 1991). In such conditions, our paradigm could be of help in the study of learning in flies, provided that the increased avoidance of the lighted vial over trials is due to learning and not to other causes, such as sensitization to the stimuli encountered by the flies during training or avoidance based on the detection of stimuli at the choice point. In our paradigm, it could be argued that the lighted arm provides a conditioned stimulus (CS—light) associated with an aversive unconditioned stimulus (US—the taste of quinine) whereas the darkened arm is not associated with an aversive stimulus. Flies would thus learn a Pavlovian association between stimuli (a CS–US association). However, since flies are naturally photopositive, it seems difficult to consider light as a CS—that is, as a neutral stimulus. This paradigm could then be assimilated to the conditioned inhibition of a reflex (as is the case with PER conditioned inhibition in fasting flies; see, e.g., Brigui et al., 1990), because flies would learn to inhibit a response toward a meaningful stimulus (i.e., light). However, the main difference between PER conditioned inhibition and the inhibition induced in our paradigm is