Improvement of spike coincidence detection with facilitating synapses
Introduction
In recent years, it has been reported that postsynaptic membrane potentials, recorded in cortical neurons, present dynamical properties which strongly depend on the presynaptic activity [1], [8]. This behaviour can be explained by considering several synaptic mechanisms such as short-term synaptic depression and facilitation. The former is well known to be responsible of several emerging complex phenomena as, for instance, cortical gain control [1], and complex switching behaviour between activity patterns in neural network models [7], [2]. This mechanism considers that synaptic buttons contain only a limited amount of neurotransmitters ready to be released. This fact turns into a situation in which the neuron is unable to transmit the incoming spikes for high presynaptic firing rate. Therefore, the resulting dynamics are highly nonlinear and strongly dependent on presynaptic activity. Moreover, synapses in cortical neurons also exhibit synaptic facilitation. This mechanism takes into account that the influx of calcium ions, through voltage-sensitive channels, favours the neurotransmitter vesicle depletion. As a consequence, facilitation is able to explain several relevant behaviour observed in real neural tissue, such as the efficient detection of bursts of action potentials (AP) [4].
In this work, we use a phenomenological model of dynamic synapses, which takes into account the two mechanisms explained above, to theoretically study their influence on the spike coincidence detection (CD). With preciseness, we compute the conditions—that is, the regions in the space of the relevant parameters of the model—in which a postsynaptic neuron can efficiently detect temporal coincidences of spikes arriving from different afferents. Our study shows that facilitation improves the detection of these correlated spikes, specially when the synapse does not have enough synaptic resources. In these conditions, depressing synapses are not able to perform well. In addition, facilitation also reveals the existence of a certain frequency value which allows the best performance for a wide range of values of the neuron firing threshold. This optimal frequency can be controlled by means of facilitation control parameters. Finally, we observe that the inclusion of the facilitation mechanism yields to a better detection of changes in the presynaptic firing rate, for certain conditions in which only depression does not perform well.
Section snippets
Model
We consider a postsynaptic neuron receiving signals from presynaptic neurons through excitatory synapses. In order to approximately model experimental data, we assume that the activity of each presynaptic neuron can be viewed as a temporal Poisson spike train with mean frequency . The state of the synapse is given by Tsodyks et al. [8]where , , are the fraction of neurotransmitters in a recovered, active and
Results
We have studied the postsynaptic response of a neuron which receives input signals from excitatory synapses. Within this population, we have considered a subset of synapses stimulated by identical spike trains. These strongly correlated afferents can be considered as a signal term. The remaining synapses receive uncorrelated spike trains constituting a noise term. We have investigated spike CD experiments by computing CD maps. In these maps, we computed the fraction of errors
Discussion
The present study shows that the inclusion of facilitation mechanisms in synapses enhances the performance of cortical neural systems in the transmission of information embedded in spike trains. Contrary to what it happens with only depression, facilitation improves CD tasks with the same amount of active neurotransmitters. Moreover, facilitation enhances good detection even for not too high correlation between signals from different presynaptic afferents. Facilitation also reveals the
Acknowledgments
This work was supported by the MEyC-FEDER project FIS2005-00791 and the Junta de Andalucía project FQM–165.
Jorge F. Mejias is currently working in his Ph.D. in the department of Matter Physics in the University of Granada (Spain). His research includes the realistic modelling of neural networks and the transmission and coding of information in neural noisy environment, as well as intracellular calcium dynamics and other biological systems.
References (8)
- et al.
Synaptic depression and cortical gain control
Science
(1997) - et al.
Effects of fast presynaptic noise in attractor neural networks
Neural Comput.
(2006) - et al.
Differential signaling via the same axon of neocortical pyramidal neurons
Proc. Natl. Acad. Sci. USA
(1998) - et al.
Differential short-term synaptic plasticity and transmission of complex spike trains: to depress or to facilitate?
Cerebral Cortex
(2000)
Cited by (0)
Jorge F. Mejias is currently working in his Ph.D. in the department of Matter Physics in the University of Granada (Spain). His research includes the realistic modelling of neural networks and the transmission and coding of information in neural noisy environment, as well as intracellular calcium dynamics and other biological systems.
Joaquin J. Torres received his Ph.D. from Granada University in 1997. Between 2001 and 2005 he has been a postdoctoral researcher (”Ramon y Cajal” contract) at department of Electromagnetism and Physics of the Matter at Granada University. Since 2005, he is an associate professor at the same department and, his current research interests are the mathematical modeling of bio-inspired neural networks and the study of the biophysics of the dynamical processes in living cells.