Search PubMed⌕ Search

PubMed · 10935461

Pattern segmentation in a binary/analog world: unsupervised learning versus memory storing.

Abstract

We discuss the problem of segmentation in pattern recognition. We adopt the model and the general approach in the landmark paper by Wang, Buhmann and von der Malsburg (Neural Computation, (1990), 2, 94-106), and expand their model in a number of ways. We review their solution to the segmentation problem in associative memory, which consists in feature binding being expressed by synchrony relations between oscillators or populations of neurons. We extend the model by introducing a law of synaptic change, which allows the network to learn by structuring itself in response to stimuli with relevant features. We discuss the problem of interference between pattern completion and the learning of new memories. We also propose a form of multiplexing of input information taking advantage of the time-structure of the neurons' response. It is based on the assessment of analog as well as of binary properties of the stimuli and provides for an enhancement of the network's processing capacity. The relevance of the results for biological systems is pointed out.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

C Lourenço, A Babloyantz, M Hougardy. 2000. Pattern segmentation in a binary/analog world: unsupervised learning versus memory storing.. https://doi.org/10.1016/s0893-6080(99)00087-8

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related citations

Muscarinic m1 and m2 receptor proteins in local circuit and projection neurons of the primate striatum: anatomical evidence for cholinergic modulation of glutamatergic prefronto-striatal pathways.

The cellular and subcellular localization of muscarinic receptor proteins m1 and m2 was examined in the neostriatum of macaque monkeys by using light and electron microscopic immunocytochemical techniques. Double-labeling immunocytochemistry revealed m1 receptors in calbindin-D28k--positive medium spiny projection neurons. Muscarinic m1 labeling was dramatically more intense in the striatal matrix compartment in juvenile monkeys but more intense in striosomes in the adult caudate, suggesting that m1 expression undergoes a developmental age-dependent change. Ultrastructurally, m1 receptors were predominantly localized in asymmetric synapse-forming spines, indicating that these spines receive extrastriatal excitatory afferents. The association of m1-positive spines with lesion-induced degenerating prefronto-striatal axon terminals demonstrated that these afferents originate in part from the prefrontal cortex. The synaptic localization of m1 in these spines indicates a role of m1 in the modulation of excitatory neurotransmission. To a lesser extent, m1 was present in symmetric synapses, where it may also modulate inhibitory neurotransmission originating from local striatal neurons or the substantia nigra. Conversely, m2/choline acetyltransferase (ChAT) double labeling revealed that m2-positive neurons corresponded to large aspiny cholinergic interneurons and ultrastructurally, that the majority of m2 labeled axons formed symmetric synapses. The remarkable segregation of the m1 and m2 receptor proteins to projection and local circuit neurons suggests a functional segregation of m1 and m2 mediated cholinergic actions in the striatum: m1 receptors modulate extrinsic glutamatergic and monoaminergic afferents and intrinsic GABAergic afferents onto projection neurons, whereas m2 receptors regulate acetylcholine release from axons of cholinergic interneurons.

Acetylcholine↗

Vesicular acetylcholine transporter in the rat nucleus accumbens shell: subcellular distribution and association with mu-opioid receptors.

Cholinergic interneurons in the nucleus accumbens shell (AcbSh) are implicated in the reinforcing behaviors that develop in response to opiates active at mu-opioid receptors (MOR). We examined the electron microscopic immunocytochemical localization of the vesicular acetylcholine transporter (VAChT) and MOR to determine the functional sites for storage and release of acetylcholine (ACh), and potential interactions involving MOR in this region of rat brain. VAChT was primarily localized to membranes of small synaptic vesicles in axon terminals. Less than 10% of the VAChT-labeled terminals were MOR-immunoreactive. In contrast, 35% of the cholinergic terminals formed symmetric or punctate synapses with dendrites showing an extrasynaptic plasmalemmal distribution of MOR. Membranes of tubulovesicles in other selective dendrites were also VAChT-labeled, and almost half of these dendrites displayed plasmalemmal MOR immunoreactivity. The VAChT-labeled dendritic tubulovesicles often apposed unlabeled axon terminals that formed symmetric synapses. Our results indicate that in the AcbSh MOR agonists can modulate the release of ACh from vesicular storage sites in axon terminals as well as in dendrites where the released ACh may serve an autoregulatory function involving inhibitory afferents. These results also suggest, however, that many of the dendrites of spiny projection neurons in the AcbSh are dually influenced by ACh and opiates active at MOR, thus providing a cellular substrate for ACh in the reinforcement of opiates.

Acetylcholine↗

Synapse-glia interactions at the mammalian neuromuscular junction.

Perisynaptic Schwann cells (PSCs) play critical roles in regulating and stabilizing nerve terminals at the mammalian neuromuscular junction (NMJ). However, although these functions are likely regulated by the synaptic properties, the interactions of PSCs with the synaptic elements are not known. Therefore, our goal was to study the interactions between mammalian PSCs in situ and the presynaptic terminals using changes in intracellular Ca(2+) as an indicator of cell activity. Motor nerve stimulation induced an increase in intracellular Ca(2+) in PSCs, and this increase was greatly reduced when transmitter release was blocked. Furthermore, local application of acetylcholine induced Ca(2+) responses that were blocked by the muscarinic antagonist atropine and mimicked by the muscarinic agonist muscarine. The nicotinic antagonist alpha-bungarotoxin had no effect on Ca(2+) responses induced by acetylcholine. Local application of the cotransmitter ATP induced Ca(2+) responses that were unaffected by the P2 antagonist suramin, whereas local application of adenosine induced Ca(2+) responses that were greatly reduced by the A1 receptor antagonist 8-cyclopentyl-1,3-dimethylxanthine (CPT). However, the presence of the A1 antagonist in the perfusate did not block responses induced by ATP. Ca(2+) responses evoked by stimulation of the motor nerve were reduced in the presence of CPT, whereas atropine almost completely abolished them. Ca(2+) responses were further reduced when both antagonists were present simultaneously. Hence, PSCs at the mammalian NMJ respond to the release of neurotransmitter induced by stimulation of the motor nerve through the activation of muscarinic and adenosine A1 receptors.

Acetylcholine↗