Pavlovian (or classical) conditioning, first reported by Pavlov (1927), is a learning process in which the pairing of a relatively neutral stimulus (the conditioned stimulus, CS) with a biologically significant stimulus (the unconditioned stimulus, US) results in the CS eliciting a response (the conditioned response, CR). This basic type of associative learning is ubiquitous among animals, including arthropods, annelids, mollusks, nematodes, and planarians (Perry et al., 2013). It contributes to the survival of animals by enabling them to find suitable food, avoid toxic food, escape predators, and detect potential mates (Domjan, 2005, Krause and Domjan, 2022, Kriete and Hollis, 2022). It also provides a solid basis for advanced forms of learning, such as social learning (Olsson et al., 2020).

Modern research in mammals, mainly utilizing rodents and primates, defines Pavlovian conditioning (or learning) as the complex process whereby experience with a conditional relationship between the CS and the US confers the CS with the ability to promote adaptive behavioral patterns (Rescorla, 1988, Fanselow and Wassum, 2016). Following the argument by Fanselow & Wassum (2016), one of the critical assumptions in the current theories of Pavlovian conditioning in mammals is that the conditioning is driven by the prediction error, i.e., the discrepancy between the predicted US and the actual US that the animal receives. Another assumption is that conditioning often forms associations between the CS and the internal representation of the US, and the expectation of the US flexibly controls the behavioral response to the CS. These theories have been developed to overcome limitations of conventional theories, which assume that the contingency or correlation between the CS and the US is sufficient to account for the occurrence of conditioning (Rescorla, 1988) and that a CR is triggered almost automatically by a CS.

Modern research in mammals has also expanded our understanding of the neural basis of Pavlovian conditioning. For instance, it has been demonstrated that specific types of dopamine neurons (DANs) in the midbrain are crucial for conditioning (Schultz, 2016), while other types are involved in executing the CR (Berridge et al., 2009, Bromberg-Martin et al., 2010). Studies exploring the functions of these DANs have provided valuable insights into the neural mechanisms of conditioning and the execution of the CR (Fanselow and Wassum, 2016, Gershman et al., 2024). However, the neural circuit mechanisms underlying current theories of Pavlovian conditioning have yet to be elucidated.

Insects are excellent models for studying the molecular, cellular, and neural circuit mechanisms of Pavlovian conditioning due to the relative simplicity of the brain organization. Research on Pavlovian conditioning in insects began with a survey of proboscis extension responses in honey bees (Apis mellifera) (Kuwabara, 1957, Takeda, 1961). After repeatedly pairing an odor (CS) with either sucrose solution (US) applied to the antennae or the proboscis, bees start to extend their proboscis when they detect the CS. Studies have shown that many features of Pavlovian conditioning in bees are similar to those in mammals (Hammer and Menzel, 1998, Giurfa and Sandoz, 2012, Menzel, 2022). Subsequent studies on fruit flies (Drosophila melanogaster), where an odor (CS) was paired with electric shock or sucrose (aversive or appetitive US), revealed that the molecular mechanisms for long-term memory in flies are conserved with mammals (Tully et al., 1994). Subsequent studies in fruit flies, utilizing powerful transgenic approaches, have highlighted that the mushroom body (MB), a key brain area in flies, plays a crucial role in Pavlovian conditioning (Owald and Waddell, 2015, Modi et al., 2020, Felsenberg, 2021). The MB is composed of three neuron types: afferent (input) neurons, intrinsic neurons (Kenyon cells), and efferent (output) neurons. When the CS and US are paired, it causes changes in the strength of synaptic transmission from Kenyon cells to MB output neurons, modifying the behavioral response to the CS. The recent development of a synaptic connectivity map of the MB in fruit flies (Li et al., 2020) presents an opportunity to elucidate the neural circuit mechanisms of Pavlovian conditioning in its entirety.

It remains unclear, however, whether modern theories of Pavlovian conditioning developed in the study of mammals apply to Pavlovian conditioning in insects. There is widespread belief that Pavlovian conditioning in insects is simpler than in mammals, because insects have only small-scale neural circuits in their brains; that is, insect brains are much smaller and their organization is much simpler than that of mammals (Mizunami et al., 1999, Mizunami et al., 2004). However, evidence is accumulating to demonstrate that information processing for Pavlovian conditioning in the insect brain is much more sophisticated than previously thought (Modi et al., 2020, Mizunami, 2021). Therefore, we need to carefully investigate the possibility that modern theories of Pavlovian learning in mammals apply to Pavlovian learning in insects.

In this review, I address the question of whether general principles of Pavlovian conditioning in insects are conserved with those in mammals. I focus on our studies on the nature and mechanisms of Pavlovian conditioning in the field cricket (Gryllus bimaculatus) that addresses the principles of Pavlovian conditioning, as well as recent studies in fruit flies and honey bees. Before entering the discussion, I briefly mention the distinction between the S-S (stimulus-stimulus) and S-R (stimulus- response) associative learning models of Pavlovian conditioning, which provides the basis for the discussion. Then, I describe the procedures of Pavlovian conditioning in crickets and discuss the roles of octopamine neurons (OANs) and dopamine neurons (DANs) in appetitive and aversive conditioning and the execution of appetitive and aversive CR, respectively, which provide insights into the principles of conditioning (or memory formation) and the execution of the CR (or memory retrieval). DA and OA are biogenic amines, with DA serving as a major neurotransmitter in both vertebrates and invertebrates. In contrast, OA is specific to invertebrates and is considered the counterpart of noradrenaline in vertebrates (Roeder, 1999). Next, I discuss the evidence suggesting that Pavlovian learning in crickets follows an error correction learning rule, and that the current value of the US controls the execution of the CR in fruit flies, crickets, and honey bees. Finally, I review the nature and mechanisms of second-order conditioning (SOC), which complements and extends the discussion on first-order Pavlovian conditioning. I suggest that the basic principles of Pavlovian conditioning in insects are conserved with those of mammals and propose that insects serve as excellent model organisms for elucidating neural network mechanisms of advanced forms of Pavlovian conditioning.