PubMed Health⌕ Search

SEARCH · PubMed Health

Results for “Reinforcement learning”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 397 records · Page 22Linked to original sources
PubMed2004

A role for ERKII in synaptic pattern selectivity on the time-scale of minutes.

Stimulus reinforcement strengthens learning. Intervals between reinforcement affect both the kind of learning that occurs and the amount of learning. Stimuli spaced by a few minutes result in more effective learning than when massed together. There are several synaptic correlates of repeated stimuli, such as different kinds of plasticity and the amplitude of synaptic change. Here we study the role of signalling pathways in the synapse on this selectivity for spaced stimuli. Using the in vitro hippocampal slice technique we monitored long-term potentiation (LTP) amplitude in CA1 for repeated 100-Hz, 1-s tetani. We observe the highest LTP levels when the inter-tetanus interval is 5-10 min. We tested biochemical activity in the slice following the same stimuli, and found that extracellular signal-regulated kinase type II (ERKII) but not CaMKII exhibits a peak at about 10 min. When calcium influx into the slice is buffered using AM-ester calcium dyes, amplitude of the physiological and biochemical response is reduced, but the timing is not shifted. We have previously used computer simulations of synaptic signalling to predict such temporal tuning from signalling pathways. In the current study we consider feedback and feedforward models that exhibit temporal tuning consistent with our experiments. We find that a model incorporating post-stimulus build-up of PKM zeta acting upstream of mitogen-activated protein kinase is sufficient to explain the observed temporal tuning. On the basis of these combined experimental and modelling results we propose that the dynamics of PKM activation and ERKII signalling may provide a mechanism for functionally important forms of synaptic pattern selectivity.

Animals↗
PubMed1994

Learning under partial reinforcement in the toad (Bufo arenarum): effects of lesions in the medial pallium.

Two experiments studied the adjustment of toads (Bufo arenarum) to partial reinforcement in a runway. In Experiment 1, two groups received 24 daily trials of either continuous reinforcement (CR) or 50% partial reinforcement (PR). Training parameters that facilitate the PR extinction effect (greater resistance to extinction after PR than CR training) in rats were selected. PR impaired performance during acquisition but had no effect on performance during extinction relative to CR. In Experiment 2, four groups were trained in a factorial design involving CR and PR, and a lesion of the medial pallium and a sham operation. Performance during acquisition was again impaired by PR, but the medial pallium lesions had no effect. The lesion, however, increased resistance to extinction after both CR and PR training. The results are discussed in relation to comparative research on learning and to the hypothesized homology of the amphibian medial pallium and the mammalian hippocampal formation.

Animals↗
PubMed2006

Dopamine D2 receptors mediate reversal learning in male C57BL/6J mice.

Dopamine is critical for directing goal-oriented behavior. We investigated dopamine D2 receptor involvement in reversal learning and reinforcement efficacy in mice lacking functional dopamine D2 receptors and their heterozygous and wild-type littermates. Mice discriminated between two odors to receive a food reinforcer. One odor signaled a reinforcer (S+); the other odor signaled no reinforcer (S-). After mice learned the S+/S- relationship, we inverted the reinforcement contingencies. The necessary number of trials to relearn the new reinforcement contingencies served as our index of reversal learning. Mice lacking functional dopamine D2 receptors repeatedly failed to inhibit previously reinforced responses during reversal trials. In a separate experiment, mice responded for reinforcers on a progressive ratio schedule of reinforcement. Mice lacking functional dopamine D2 receptors earned significantly fewer reinforcers than did heterozygous mice. Our results suggest that dopamine D2 receptors regulate reversal learning and influence the reinforcing efficacy of natural rewards.

Analysis of Variance↗
PubMed1993

The effects of atrial natriuretic peptide on food-reinforced conditioning in rats. Interactions with neurotransmitters.

The effects of two doses of rat atrial natriuretic peptide (TANP-1-28), 200 and 500 ng, on 6-day acquisition and extinction of food-reinforced conditional learning (conditional stimulus: light) were studied in rats following administration into the lateral cerebroventricle. With the higher dose, there was a tendency for facilitated acquisition and significantly delayed extinction of the positively reinforced learning task. In order to clarify whether the effect of the peptide is obtained through the involvement of neurotransmitters, the experimental animals were pretreated with different receptor blockers in selected doses that did not influence the behavioral test. Haloperidol, atropine, phenoxybenzamine, and propranolol blocked the action of ANP on extinction of the food-reinforced conditioning, whereas naloxone, bicuculline, and methysergide were ineffective. The results suggest that ANP might be considered a modulating agent in a positively reinforced conditional learning task, and that its action might involve dopaminergic, cholinergic, and alpha- and beta-adrenergic mechanisms.

Animals↗