Details

Autor(en) / Beteiligte

• Buatois, Alexis
• Laroche, Lou
• Lafon, Gregory
• Avarguès‐Weber, Aurore
• Giurfa, Martin

Titel

Higher‐order discrimination learning by honeybees in a virtual environment

Ist Teil von

• The European journal of neuroscience, 2020-01, Vol.51 (2), p.681-694

Ort / Verlag

France: Wiley Subscription Services, Inc

Erscheinungsjahr

2020

Link zum Volltext

Quelle

Psychology & Behavioral Sciences Collection

Beschreibungen/Notizen

• Non‐elemental learning constitutes a cognitive challenge because events to be learned are usually ambiguous in terms of reinforcement outcome, contrary to elemental learning, which relies on unambiguous associations. Negative patterning (NP) constitutes a paradigmatic case of non‐elemental learning, as subjects have to learn that single elements are reinforced while their simultaneous presentation is not reinforced (A+, B+ vs. AB−). Solving NP requires treating AB as being different from the linear sum of its components in order to overcome the ambiguity of stimulus reinforcement (i.e. A and B are as often reinforced as not reinforced). The honeybee is currently the only insect mastering NP as shown by studies restricted mainly to the olfactory domain. Here, we tested the bees' capacity to solve a NP discrimination in the visual domain and used to this end a virtual reality (VR) environment in which a tethered bee walking stationary on a spherical treadmill faces visual stimuli projected on a semicircular screen. We show that bees learn a composite grating made of alternated green and blue bars in an elemental way, and generalize their response to both a blue and a green grating. Yet, after NP training, one‐quarter of the bees inhibited elemental processing and responded significantly more to the single‐coloured gratings than to the composite grating. Alternative strategies were used by the other bees, which achieved partial NP learning. These results offer attractive perspectives to study different forms of visual learning in a controlled VR environment, and dissect their underlying mechanisms. Negative patterning is a discrimination problem in which subjects have to learn that a non‐rewarded compound AB is an entity different from the sum of its rewarded constituents A and B (i.e. A+, B+ vs. AB−). We studied the capacity of honeybees to solve a negative patterning discrimination in the visual domain, by training them with coloured gratings (A: green; B: blue; and AB: green‐blue) in a virtual reality environment. Bees perceived the elemental gratings A and B within the composite grating AB and generalized highly between them; yet, they decreased significantly generalization when trained under a negative patterning regime and preferred A and/or B to AB. The virtual reality conditions in which this sophisticated learning took place open new perspectives for accessing the neural bases of this performance.