Many brain faculties depend on the ability to detect and encode associations and correlations among events. Coincidence detectors are essential for this ability. The most straightforward type of coincidence detector senses the occurrence of two events at the same time (definition 1). However, what is meant by 'at the same time' depends on the scientific discipline and on the "level of analysis. Simultaneity in particle physics has a different meaning than in physiology and psychology. Hence, although an ideal coincidence detector responds only when the time difference between events is t = 0, in practice, a response is emitted to t = 0 ±At. It is safe to say that in the context of neuroscience, two events can be said to occur 'at the same time' if that time is in the order of magnitude of milliseconds (less then a millisecond to a few milliseconds in synaptic events, e.g. Markram et al. 1997; < 100 ms in perception and cognition, e.g. Thorpe et al. 1996; van Turennout et al. 1998; "percept). Note that if At> 0, the order of events may also make a difference, i.e. whether i arrives before j or vice versa (see "classical conditioning and below). Another type of coincidence detector generates the desired output only after receiving two or more "stimuli in a number of steps within a defined, commonly short time window ('graded coincidence detector'; definition 2). Still another type of integrating device, which is also sometimes referred to as a 'coincidence detector', is characterized by more relaxed temporal constraints, and in this case 'coincidence' refers to occurrence at the same place rather than at the same time (definition 2). An additional point that requires clarification is what is meant by 'device'. In the context of the present discussion, a device is either a macromolecule, a "synapse, a nerve cell, or a neuronal circuit.
The majority of experimental evidence for the involvement of identified coincidence detectors in learning and memory, relates to the molecular and synaptic level. An example is provided by the enzyme "calcium/calmodulin-sensitive adenylyl cyclase, that generates the second messenger cyclic adenosine monophosphate (cAMP; "intracellular signal transduction cascades, *CREB). In the circuit that subserves the conditioning of defensive reflexes in "Aplysia, a modu-latory "neurotransmitter is assumed to encode the unconditioned stimulus (US), and calcium the conditioned stimulus (CS). Both activate adenylyl cyclase, culminating in intracellular cascades that subserve memory of the modified reflex (see also "Drosophila, "immediate early genes). Interestingly, the activation of the cyclase was reported to be larger when the calcium pulse immediately preceded the transmitter (Yovell and Abrams 1992). This led to the proposal that temporal asymmetry in the activation of the cyclase underlies temporal asymmetry of the CS-US pairing in the behaving organism (ibid., also Abrams and Kandel 1988, and "classical conditioning).
Another molecular coincidence detector with a proposed role in "plasticity and learning is the "gluta-matergic N-methyl-d-aspartate (NMDA) "receptor (NMDAR; Seeburg et al. 1995). Here, the receptor-"channel complex detects coincidence of presynaptic activity (glutamate, which activates the receptor) and postsynaptic electrical activity (depolarization, which removes a magnesium block from the channel, Figure 16). Convergence in this case is permitted within tens to hundreds of milliseconds. Furthermore, at least
Fig. 16 The W-methyl-d-aspartate-type of the *receptor for the 'neurotransmitter 'glutamate is an example of a molecular coincidence detector that takes part in cellular 'plasticity mechanisms, such as 'long-term potentiation, which are assumed to subserve learning and memory. In the resting state (left panel in this highly simplified scheme), the extracellular (Out) binding site (B) for glutamate (Glu) is vacant and the 'ion channel, which can permit influx of 'calcium (o) into the neuron, is blocked by magnesium (•). Electrical activity that results in postsynaptic depolarization can relieve the magnesium block, but this by itself is not sufficient to allow a significant calcium influx. Occupation of the glutamate binding site by the transmitter is also insufficient to open the channel. Only coincident activity of glutamate and depolarization (right panel) leads to calcium influx (heavy arrows, Yes), which may culminate in plastic changes in the neuron. Cyt, cytoskeleton, with which the intracellular (In) portion of the receptor interacts; X, binding sites for other molecular 'stimuli.
Yes one additional type of input, glycine, and possibly more, are required for optimal activation. The relevance of the coincidence detection properties of NMDAR to identified behaviours has not yet been determined. Other macromolecules have been shown to function as coincidence detectors (Bourne and Nicoll 1993; Caroll et al. 1995). Notable among them are transcriptional regulation elements, including transcription factors and promoters (Impey et al. 1994; Janknecht et al. 1995; Deisseroth et al. 1996; "CREB, "immediate early genes, "protein synthesis). Some cellular scaffold proteins (e.g. such that mediate "protein kinase activation; Whitmarsh et al. 1998) can also be considered as signal integrators (definition 2). Two major conclusions can be drawn from the data. First, coincidence detectors exist at multiple tiers of intracel-lular signal transduction cascades, from the membrane downstream to the cell nucleus, hence yielding many permutations in the interaction of stimuli and as a consequence, high response specificity. Second, 'coincidence' should be considered as an intracellular representational code, which is embodied in intracellu-lar molecular networks ("reduction).
Coincidence detection has been described in several circuits that process and encode sensory information
(Sullivan and Konishi 1986; Hopfield 1995; Edwards et al. 1998), although the relevance of these findings at the circuit level to learning and memory, if at all, has not yet been determined. One of the best known examples is coincidence detection in the circuit that encodes interaural phase differences in the brainstem of birds and mammals, and permits very precise localization of sounds. In this system, coincidence detection combines with the use of neuronal delay lines, which carry the auditory information from each ear separately, to convert a code that is based on the time of arrival of neuronal firing from the ear ('time code') into a code that is based on the location of the firing neurons in the brain ('place code'; Sullivan and Konishi 1986; also Agmon-Snir et al. 1998; "map). In general, discussion of coincidence detection at the circuit level relates to two classes of issues: on the one hand, the spontaneous and the evoked responses of the neurons in the circuit; on the other, the putative "representational codes used by the circuit (de No 1938; Hebb 1949; von der Malsburg 1987; Abeles 1991; Hopfield 1995;Konig et al. 1996; Dan et al. 1998). The neurons in the circuit integrate incoming synaptic potentials before they fire a 'spike'. The kinetics of this integration is a parameter that determines whether the neuron can serve only as an integrator (definition 2) or also as a bona fide temporal coincidence detector (definition 1). Only if the integration time interval is shorter than the interspike time interval, can the neuron act as a coincidence detector for the incoming spikes. The synchronized information, which is then relayed to other neurons, might encode the coherency of a sensory "percept or a mental concept (Konig et al. 1996).
All in all, it is difficult to envisage how without the evolution of coincidence detectors, be they molecules, neurons, or circuits, our brain would have had the capability to detect correlations and causal relations in the external world, and "bind together meaningful narratives in the internal world.
Selected associations: Algorithm, Cell assembly, Internal representation, Intracellular signal transduction cascade
Was this article helpful?