Midbrain Dopaminergic Neurons and Striatal Cholinergic Interneurons Encode the Difference between Reward and Aversive Events at Different Epochs of Probabilistic Classical Conditioning Trials
Mati Joshua, Avital Adler, Rea Mitelman, Eilon Vaadia, and Hagai Bergman
The Journal of Neuroscience (2008)
Doi: 10.1523/JNEUROSCI.3839-08.2008
Brief summary: Joshua and colleagues found that dopamine (DA) neurons could effectively differentiate the motivational value of predictive cues and outcomes, while the tonically active neurons (TANs, putative cholinergic interneurons, ChIs) were better at the detection of the omission of the expected outcomes (both reward and aversive air-puff).
Previous study from this lab had shown that TANs concomitantly and oppositely responded to motivationally significant external stimuli (Coincident but Distinct Messages of Midbrain Dopamine and Striatal Tonically Active Neurons, Neuron, 2004, shared in 2021-07-15). In that study, they only used reward predicting cues. How about the responses of TANs to the cues predicting aversive outcome or to the aversive outcome per se? In the current study, they utilized three groups of pictures to predict reward, air-puff and neutral auditory clicks, respectively, at different probabilities, and found that DA neurons were good at discriminating both the type and reward probability of these cues both during cue presentation and outcome periods, while TANs responded to these cues un-selectively during cue presentation period and with a similar pause-rebound pattern to the aversive and appetitive outcomes (with a shorter latency to the aversive air-puff). The results suggested that DA neurons were more suited for supporting stimuli-reward conditioning, while the TANs may be important for the detection of the general salience of external stimuli. Although DA neurons were widely believed to detect the reward prediction errors (the discrepancy between actual outcome and the predicted), unexpectedly, TANs responded to the omission of the outcomes (both to the reward and the aversive air-puff un-selectively) with stronger suppression comparing to that of the DA neurons. This observation strongly implied that TANs/ChIs may be more important for the outcome-omission driven learning.
The observation of ChIs in detecting reward omission is very inspiring! It suggests that the suppression of the inputs (from cortices or thalamus) to striatum may facilitate the restraint of response under the no-go behavioral conditions via the inhibition of ChIs. And it also implicitly implies that, at the physiological conditions, the excitatory inputs to ChIs endow the striatum to support more explorative behaviors, even though some behaviors can not lead to rewards immediately (so they are energy consuming). Indeed, the prediction had been partially verified by a later study (Enhanced flexibility of place discrimination learning by targeting striatal cholinergic interneurons, Nature communications, 2014, shared in 2021–7-07). In that study, the researchers found that suppression of the thalamostriatal inputs could significantly improve the learning of place reversal learning. So can we specifically suppress any cortical inputs to ChIs to facilitate the learning processes (by refraining the prepotent responses in the no-go conditions) when the omission of reward happens? It’s an important conceptual gap, which is deserved to be verified experimentally.
