Biomedical Engineering Reference
In-Depth Information
of the migrating cells reached the target and developed functional connec-
tions to other neurons in HVC [Burd and Nottebohm 1985]. To make the
picture even more suggestive, many of these new neurons developed projec-
tions to RA [Nordeen and Nordeen 1988]. As Catchpole and Slater point out
[Catchpole and Slater 1995], there are species in which neurogenesis is not
accompanied by changes in the song, as is the case with the zebra finch.
Unquestionably, there is an important mechanism to be revealed relating
neurogenesis and learning. However, it might not be the only neural process
involved in learning.
As we have discussed, there is an indirect connection between the HVC
and RA nuclei. It is known as the anterior forebrain pathway, and does not
participate in the production of song, but plays a crucial role in learning.
The output of this pathway is the lateral magnocellular nucleus of the an-
terior neostriatum. A series of experiments has suggested that this nucleus
responds if some error is detected [Bottjer 2002]. Individual RA neurons re-
ceive inputs from both lMAN and HVC, which is consistent with the picture
that experience-related lMAN activity facilitates certain HVC-RA synapses,
helping to build the neural architecture necessary to produce the adult song.
According to this picture, a sequence of bursts generated at an RA-
projecting HVC neuron will induce some activity in RA, and also eventu-
ally induce an activity in lMAN that will lead to either the potentiation
or the depression of the connection. This signal, however, requires a time
for processing through the AFP, which has been estimated as approximately
40 ms [Kimpo et al. 2003]. Abarbanel and coworkers [Abarbanel et al. 2004b]
presented a biophysical model for the heterosynaptic facilitation mechanism
described above, assuming a sparse activity in the RA-projecting HVC neu-
rons like that reported in the literature [Hahnloser et al. 2002]. Abarbanel
and coworkers showed that long term potentiation (LTP) is possible pre-
cisely if the delay between the presynaptic activity of the HVC connection
and the presynaptic activity arriving from lMAN is less than or equal to
40 ms. Therefore, the estimated processing times of the AFP allow the sys-
tem to operate in a regime in which either LTP or long term depression (LTD)
can be easily achieved through minimal dynamical changes. The basic idea
is that a change in calcium concentration in an RA cell can lead to changes
in the conductance of the HVC-RA synapses, and that this change in con-
centration can be induced by the arrival of bursts of action potentials from
HVC and lMAN, separated by specific intervals of time (∆
T
). The model is
computational: it describes the time evolution of the membrane potential of
an RA neuron by a conductance model
C
M
dV
dt
V
(
t
)) +
I
NMDA
HV C
+
I
AMP A
HV C
+
I
NMDA
=
g
L
(
E
L
−
lMAN
,
(8.29)
where the “leak” current drives the neuron to
V
=
E
L
in the absence of
signaling from HVC or lMAN,
C
M
is the membrane capacitance per unit
area, and the currents are synaptic currents induced by activity in lMAN or
