Home
This Title All WIREs
WIREs RSS Feed
How to cite this WIREs title:
WIREs Cogn Sci
Impact Factor: 2.88

Invertebrate learning and cognition: relating phenomena to neural substrate

Full article on Wiley Online Library:   HTML PDF

Can't access this content? Tell your librarian.

Diverse invertebrate species have been used for studies of learning and comparative cognition. Although we have gained invaluable information from this, in this study we argue that our approach to comparative learning research is rather deficient. Generally invertebrate learning research has focused mainly on arthropods, and most of that within the Hymenoptera and Diptera. Any true comparative analysis of the distribution of comparative cognitive abilities across phyla is hampered by this bias, and more fundamentally by a reporting bias toward positive results. To understand the limits of learning and cognition for a species, knowing what animals cannot do is at least as important as reporting what they can. Finally, much more effort needs to be focused on the neurobiological analysis of different types of learning to truly understand the differences and similarities of learning types. In this review, we first give a brief overview of the various forms of learning in invertebrates. We also suggest areas where further study is needed for a more comparative understanding of learning. Finally, using what is known of learning in honeybees and the well‐studied honeybee brain, we present a model of how various complex forms of learning may be accounted for with the same neural circuitry required for so‐called simple learning types. At the neurobiological level, different learning phenomena are unlikely to be independent, and without considering this it is very difficult to correctly interpret the phylogenetic distribution of learning and cognitive abilities. WIREs Cogn Sci 2013, 4:561–582. doi: 10.1002/wcs.1248 This article is categorized under: Psychology > Comparative Psychology Psychology > Learning
Honeybee brain schematic highlighting the olfactory system. When odors are sensed by the antennae (An) odorant information is transferred via the antennal nerve to the antennal lobe (AL). Within the AL olfactory neurons synapse with projection neurons (PN) and interneurons within spherical glomeruli. PNs convey odor information as a cross‐fiber code to the lateral horn and the calyces of the mushroom bodies (MB) where PNs synapse with the intrinsic neurons of the MBs, the Kenyon cells. Kenyon cells output via the α‐lobe (α) and β‐lobe (β). In these regions, Kenyon cells synapse with extrinsic neurons (EN) which connect the MBs with the lateral horn (LH) of the protocerebral lobe (PL) and other regions. Centrifugal neurons (CN) also send feedback information to the antennal lobes.
[ Normal View | Magnified View ]
This simple conceptual model of how different forms of learning can be supported by the same neural system is developed from a series of formal mathematical models of the bee olfactory learning pathway. Changes in the activity of extrinsic neurons (ENs) to odors drive both nonassociative and associative learning. The panels show the populations of ENs responding above threshold to a given odor, with neurons colored white activating proboscis extension, and neurons colored black activating proboscis retraction. Reciprocal lateral inhibition in this population makes the behavioral outcome of the inhibitory interactions into a ‘winner takes all’ situation, in which the behavioral response is determined by whether the extension or retraction groups are numerically dominant in the active EN population. The default response of ENs to an odor is that ‘retraction’ neurons dominate. Consequently bees do not normally extend their proboscis to a novel odor. (a). Single odor conditioning. After association of the odor with sucrose reward, the pattern of activation of ENs changes such that additional neurons of the proboscis extension group are recruited. The ‘winner takes all’ feature resulting from lateral inhibition results in, post‐training, bees extending their proboscis to the odor. Associative conditioning with an aversive stimulus also changes the response of ENs to the odor such that fewer ‘extension’ neurons are activated and more ‘retraction’ neurons are activated. Repeated presentation of an odor alone with no consequence (habituation) causes similar, but less extreme, changes as training with an aversive stimulus. Latent inhibition describes the observation that appetitive conditioning with reward takes longer if animals have been first preexposed to the odor (CS) alone. This outcome is a natural consequence of the model because following preexposure of odor alone the number of extension group ENs activated by the odor is reduced, therefore requiring more conditioning trials with reward to bring the extension response to dominance. For similar reasons, our model explains why the process of reversal learning (pairing an odor with reward after first pairing an odor with an aversive stimulus) is slower than simple appetitive conditioning: a greater shift in the pattern of active ENs is needed to bring proboscis extension to an odor to dominance in a reversal learning paradigm than in simple reward conditioning. (b) Discriminating two dissimilar odors. Odors X and Y are perceptually distinct and activate fully distinct populations of Kenyon cells. The compact nature of odor coding in the ENs results in some degree of overlap of EN populations responding to initial presentations of X and Y, meaning some neurons will respond to both odors. On initial odor presentation the retraction group dominates the response to both odors. Following differential conditioning of X with sucrose reward (CS+) and Y with an aversive stimulus (CS−) the populations of ENs responding to the two odors diverge. As a result of Hebbian processes extension group neurons are recruited to odor X, such that the extension group now dominates the EN response. The EN population responding to odor Y also shifts such that the retraction group is now even more dominant. (c) Discriminating two similar odors. Odors X and Z are perceptually similar and activate overlapping Kenyon cell populations. As a result, a significant number of extrinsic neurons will initially respond to both odors. Following differential conditioning to X (CS+) and Z (CS−), the extrinsic neuron populations activated by the two odors diverge, but cannot become completely distinct because of the high degree of odor similarity. For the extension EN neurons responding to both odors, strengthening of connections as a consequence of training with X is counteracted by training with Z. For the retraction EN neurons responding to both odors strengthening of connections as a consequence of training with Z is counteracted by training with X. These antagonistic processes will slow learning: it takes more training for the extension group to dominate the response to the CS+ odor X, and after training the extension group will dominate the response to X less than when following training with distinct CS+ and CS− in (b). This process also explains peak shift. After differential conditioning of similar stimuli, the maximal response occurs not to the CS+ odor X, but to a new odor (W) that is similar to X but more distinct from the CS− odor (Z). ENs responding to W are expected to overlap significantly with ENs responding to X, but overlap less with extrinsic neurons responding to Z. As a consequence of training with X and Z, the EN response to W may be more dominated by the extension group than the response to X. Consistent with empirical results, this interpretation suggests the occurrence of peak shift if X and Z (CS+ and CS−) are very similar. The distinct X and Y described in (b) would not produce peak shift.
[ Normal View | Magnified View ]
Schematic of the honeybee olfactory learning pathway. Neuron cell bodies are shown as spheres, diamonds mark points of synaptic contact between projections. Lights gray neurons are active, dark gray inactive. The diagram highlights the organizational structure of the honeybee olfactory pathway as three serial connection matrices (1) between the olfactory receptor neurons (ORN) and the projection neurons (PN) within the glomeruli (G), (2) between the PN and the Kenyon cells (KC) within the calyx of the mushroom bodies (MB) and between the KC and extrinsic neurons (EN) within the lobes of the MB. The ENs output to premotor regions, including the lateral horn of the protocerebral lobe (PL), considered important for organizing the conditioned response. Odor information is encoded as a cross‐fiber code within the PNs and KCs. Reward and punishment cause release of neuromodulators that change the pattern of activity within the KCs and ENs as a neural engram of the learning process.
[ Normal View | Magnified View ]

Browse by Topic

Psychology > Learning
Psychology > Comparative Psychology

Access to this WIREs title is by subscription only.

Recommend to Your
Librarian Now!

The latest WIREs articles in your inbox

Sign Up for Article Alerts