You are on page 1of 10

Available online at www.sciencedirect.

com

ScienceDirect

Synaptic plasticity in dendrites: complications and


coping strategies
Bartlett W Mel1, Jackie Schiller2 and Panayiota Poirazi3

The elaborate morphology, nonlinear membrane mechanisms by identifying both the variables and the interactions
and spatiotemporally varying synaptic activation patterns of between them that determine how the connection
dendrites complicate the expression, compartmentalization between two neurons should change.
and modulation of synaptic plasticity. To grapple with this
complexity, we start with the observation that neurons in Most real neurons do not physically resemble Hebb’s
different brain areas face markedly different learning problems, neurons A and B, however. Rather, they are dominated by
and dendrites of different neuron types contribute to the cell’s dendritic trees having a wide variety of shapes and sizes
input–output function in markedly different ways. By [2–5]. Dendrites complicate the discussion of synaptic
committing to specific assumptions regarding a neuron’s plasticity since the number of variables that can influence
learning problem and its input–output function, specific the direction and magnitude of changes at a given synapse
inferences can be drawn regarding the synaptic plasticity is radically increased compared to Hebb’s simple picture
mechanisms and outcomes that we ‘ought’ to expect for that [155]. Perhaps the simplest way dendrites complicate
neuron. Exploiting this assumption-driven approach can help neural plasticity occurs when different plasticity rules
both in interpreting existing experimental data and designing apply to different dendritic subtrees (e.g. apical vs. basal)
future experiments aimed at understanding the brain’s myriad within the same neuron [6,7–11], or at a finer scale, to
learning processes. different subregions within the same dendrite [12,13]
(Figure 1b). Dendrites introduce more profound compli-
cations when we consider that the voltage signals govern-
Addresses
1 ing plasticity at a given dendritic location will be influ-
Biomedical Engineering Department and Neuroscience Graduate
Program, University of Southern California, Los Angeles, CA 90089, enced by a multitude of factors. First, local passive cable
United States properties of dendrites and spines can vary dramatically
2
Department of Physiology, The Rappaport Faculty of Medicine and across the dendritic arbor [14–22], influencing the size
Research Institute, Technion-Israel Institute of Technology, Haifa 35254, and time course of the local voltage responses caused by
Israel
3
Institute of Molecular Biology and Biotechnology (IMBB), Foundation
synaptic conductance changes. The voltage signals felt by
for Research and Technology-Hellas (FORTH), Heraklion, Crete, Greece a synapse are also influenced by a multitude of excitatory
and inhibitory inputs activated elsewhere in the dendritic
Corresponding authors: Mel, Bartlett W (mel@usc.edu), arbor, whose combined effects at any given location
Poirazi, Panayiota (poirazi@imbb.forth.gr)
depend in complicated ways on distance travelled [23–
25], intervening dendritic diameters and branching struc-
Current Opinion in Neurobiology 2017, 43:177–186 ture [14,15,26,27], and boosting or suppression by volt-
This review comes from a themed issue on Neurobiology of learning age-dependent membrane mechanisms [28–33]—which
and plasticity can themselves undergo activity-dependent changes that
Edited by Leslie Griffith and Tim Vogels
modify dendritic excitability [6,34–43]. Plasticity out-
comes can also depend on the relative timing of synaptic
For a complete overview see the Issue and the Editorial
inputs and back-propagating somatic action potentials
Available online 25th April 2017 [8,9,44–47], whose penetration into the dendritic arbor
http://dx.doi.org/10.1016/j.conb.2017.03.012 can be time, morphology and state-dependent [10,39,48–
0959-4388/ã 2017 Elsevier Ltd. All rights reserved. 51].

Synaptic plasticity can also be modulated by pathways


that target-specific subregions of a dendritic tree [52], or
even specific parts of a dendritic branch [13]. Inhibition,
for example, even if applied uniformly across a dendritic
Introduction tree, can implicitly promote compartmentalization of
When Donald Hebb proposed his famous learning rule plasticity by raising the threshold for plasticity induction,
[1], he referred to the very simple situation in which so that only the most strongly activated dendritic com-
neuron A participates repeatedly in the firing of neuron B, partments can participate in learning. This effect by itself
leading to a stronger connection between them can lead to differentiation of the functions assigned to
(Figure 1a). Hebb’s rule may be the single most influen- different dendrites [53,54]. Dendrite-wide threshold
tial idea in the field of neural plasticity; it has affected the elevation likely does not fully capture the extent to which
thinking of generations of theorists and experimentalists inhibition can spatially restrict plasticity, however.

www.sciencedirect.com Current Opinion in Neurobiology 2017, 43:177–186


178 Neurobiology of learning and plasticity

Figure 1

(a) Hebb’s scenario (circa 1949)

A B
w

(b)
excitatory

inhibitory

firing

depolarized

silent

distal proximal

Current Opinion in Neurobiology

Synaptic plasticity within complex dendritic trees.


(a) The basic synaptic plasticity scenario suggested by Donald Hebb in 1949. The connection (w) between two neurons A and B increases if the
activity of neuron A repeatedly causes the firing of neuron B.
(b) Synaptic plasticity rules are potentially influenced in myriad ways by electrical and chemical signaling in dendritic trees. Pyramidal neurons, for
example, are organized into distinct anatomical compartments, including the apical tuft (yellow shading), the main apical tree (blue shading) and
the basal tree (green shading), within which synaptic plasticity rules can differ. Moreover, individual dendrites within these compartments can be
found in various activity states (e.g. silent, depolarized, firing) and these activity states along with synaptic input at both excitatory and inhibitory
synapses influence the production/diffusion of plasticity related molecules (green, yellow) that are necessary for synaptic plasticity. Finally, the
location of synaptic input, whether distal or proximal within a dendritic branch also influences synaptic plasticity as different rules may operate at
distinct locations within a branch.

Recent evidence suggests that interneurons can inhibit [23,24,33,60–62] adding to the list of mechanisms that
certain dendrites within a neuron but not others [55–57], can influence synaptic plasticity outcomes at different
can target-specific parts of specific classes of dendrites dendritic places and times within the same neuron.
(e.g. proximal parts of terminal basal dendrites [58], or
can even target a restricted part of a single branch within Dendrites not only influence the communication of
a neuron’s dendritic arbor [59]. This scale of wiring voltage signals, but also chemical signals [63] that can
specificity of inhibition is known from modeling studies introduce distance and time dependence to local
to have profound effects on dendritic integration plasticity rules, including Ras [64], Arc [65,66], numerous

Current Opinion in Neurobiology 2017, 43:177–186 www.sciencedirect.com


Synaptic plasticity in dendrites: complications and coping strategies Mel, Schiller and Poirazi 179

plasticity-related proteins (PRPs) [67–70], as well as faces. In yet another example, knowing that a neuron is
mobilization and recruitment of receptors and channels involved in the storage of episodic memories, which by
such as AMPA, Kv4.2, and HCN channels [71–75]. definition occur only once, leads to the expectation that
any synaptic changes triggered by those experiences must
Finally, dendrites create the opportunity for ‘wiring plas- be essentially full and immediate [100,101].
ticity’, allowing a neuron to tap into huge capacity gains
[76–79], information binding [80], temporal linking of Three learning scenarios
memory engrams [81], detection of temporally corre- To explore these ideas further, we consider three cases
lated signals [82], and other benefits [83]. The elaborate where a neuron finds itself in a different learning and/or
array of biological mechanisms controlling the formation, developmental scenario, and explore the requirements,
stabilization and elimination of synaptic connections expectations, and counter-expectations regarding the
between axons and dendrites must therefore be consid- neuron’s dendritic readout mechanism(s) and plasticity
ered as an intrinsic part of the overall synaptic learning rules. Some of the inferences drawn are speculative, but
scheme [84–89]. nonetheless illustrate the way assumptions about the
learning problem can powerfully influence our expecta-
In summary, in comparison to Hebb’s classical A ! B tions as to how the neuron should operate, and how its
plasticity scenario, the mechanisms that collectively parameters should change during learning. Given this
determine when, where, how much, in what direction, review pays particular attention to the role dendrites play
and with what effect plasticity occurs at a given site within in shaping plasticity rules and outcomes, as a reference
a dendritic tree are manifold, heterogeneous, and com- case we begin with a scenario where sophisticated den-
plex, and we have far to go before reaching either a solid dritic processing is presumably not needed to learn or
or a comprehensive understanding of the impact of den- perform the task (Scenario 1). We then move on to
drites on the plasticity of neural tissue. consider two scenarios where subunitized dendritic com-
putation may be helpful or necessary, but differing in
The importance of learning problems and other important respects, such as whether the learning
readout mechanisms process primarily involves weight vs. wiring plasticity, and
A major thesis of this review is that, to help impose order whether synaptic changes are expected to be binary or
on the complex data field relating to local plasticity rules graded, among several other differences (see Table 1).
in dendrites, we can exploit the tight relationship that
exists between: Scenario 1: A neuron that participates in a large number of
different sensory-motor tasks, where each task requires that the
 the high-level learning (or developmental) problem cell learn a different, smooth, time-varying function that
faced by a neuron, depends on a vast number of input variables.
 the neuron’s input–output function, and
 the plasticity rules whose job is to set the neuron’s Consider a cerebellar Purkinje cell that controls a shoul-
operating parameters. der muscle or muscle synergy that participates in a diverse
set of sensory-guided tasks such as throwing, carrying,
chopping, digging, running, fighting, dressing, and a
In particular, knowledge gained about any of the three multitude other tasks that cannot be anticipated – and
can create useful expectations about the others. For thus cannot be hardwired – by developmental processes
example, it would be surprising to discover in some brain or evolution. The neuron must therefore be trained to
area an elaborate biological infrastructure that encourages produce a multitude of different time-varying continuous
correlated groups of axons to seek out and form clusters of valued functions that encode each skill. In the course of
synaptic contacts within the dendrites of their target each behavior, the neuron’s firing rate depends differ-
neurons if the target neurons’ input–output function ently on a vast number of somatosensory, proprioceptive,
ignored such clustering. This is a case where knowing vestibular, visual, motor efference copy, and contextual
that the plasticity rule creates clusters of inputs leads us to inputs. Given the neuron’s input space contains so many
expect that the neuron’s readout mechanism should be dimensions (presumably by design—see Refs. [102,103]),
sensitive to clusters of inputs. It would likewise be the winding, often cyclic input-space trajectories tra-
surprising if a neuron was found to contain separately versed by the neuron will scarcely overlap between tasks
thresholded dendritic subunits [18,31,39,90,91], and was [104]. As such, the learning problem faced by a neuron
therefore highly sensitive to clustered patterns of synaptic can be caricatured as an ultra high-dimensional classifica-
activation [31,76,92–95], but those functional subunits tion problem in which the neuron learns to ‘recognize’,
played no role in solving the neuron’s high-level learning through repeated training, a large number of randomly
problem [96,97,98,99]. In this case, information about scattered locations in a vast input space where its firing
the neuron’s readout mechanism creates expectations rate must be elevated or depressed relative to baseline.
about the high-level learning problem(s) that the neuron To the extent that this accurately describes the Purkinje

www.sciencedirect.com Current Opinion in Neurobiology 2017, 43:177–186


180 Neurobiology of learning and plasticity

Table 1

Three different learning/developmental problems and their plasticity characteristics

Sensory-motor skill Online familiarity-based Development of a ‘complex cell’


learning recognition memory receptive field
Putative cell type/location Purkinje cell in cerebellum Pyramidal cell in perirhinal cortex Pyramidal cell in early visual cortex
Preferred size of learning unit Whole neuron Dendritic subunit Dendritic subunit
Primary plasticity mode Weight plasticity Weight plasticity Wiring plasticity
Connectivity to dendritic tree Random Random Early: random
Late: structured
Subunit internal structure n/a Flat Spatially sensitive
Synaptic weights Graded Binary Graded
Training Repeated One shot Repeated
Supervisory signal Cell wide None Early: somatic bAPs
Late: dendritic potentials
role of LTP Learning in both cases Learning To stabilize good connections
role of LTD Forgetting To eliminate bad connections
Synaptic meta states helpful? ? ‘age’ ?
(since last LTP event)

cell’s learning scenario, from an engineering perspective A memory of this type, sometimes called a ‘palimpsest’,
the simplest design strategy is to construct a single-layer receives a continuous sequence of patterned inputs, and
‘Perceptron’4 [105] with modifiable synapses from all must store a trace of each pattern based on a single
potentially relevant inputs for all possible tasks combined presentation, while causing as little disruption to previ-
through a simple weighted sum [102,103,106,107,151, ously stored memories as possible. Performance is
152], see Figure 2a . [Given that a Purkinje cell receives assessed by measuring the system’s ability to distinguish
on the order of 250 000 synaptic contacts, its capacity to familiar patterns from novel distractors as a function of the
‘label’ input patterns as described above is immense—on time elapsed since the original presentation [114–120].
the order of 250 000 [108,109]]. Following from this Familiarity-based recognition is thought to be centered in
model, we expect learning to be implemented by graded, the perirhinal cortex within the medial temporal lobe
bidirectional, long-term-stable synaptic changes medi- [121,122]. Even when axons are wired to dendrites at
ated by conventional LTP and LTD processes [110], random (as in Scenario 1), the existence of separately
and that the synaptic weight values will conform to a thresholded dendritic subunits, which are devoted to
predictable steady-state probability distribution [111]. recognizing partial patterns, has been shown to signifi-
Given that a Perceptron is essentially a linear device, cantly boost online storage capacity [120,123]. The
we do not expect to observe location-dependent nonlin- expectation in this scenario, therefore, is that the plastic-
ear interactions between parallel fiber inputs within ity rule should operate at the level of the individual
Purkinje cell dendrites, as they are not needed either dendrite, and should involve both cooperative and
during learning or readout [151,152]. Furthermore, competitive interactions among nearby synapses in each
given both the nature of the learning problem and the dendrite in which learning has been induced
highly constrained physical architecture of the parallel [81,85,87,124,125]. The act of storing a pattern should
fiber ! Purkinje cell interface, we do not expect to involve one- shot LTP [12,101,120], whereas LTD
uncover biological machinery that supports learning- should be responsible for ‘forgetting’ at the end of the
driven structural (wiring) plasticity between axons and memory’s lifetime [126,127]. Unlike the Perceptron
dendrites, as evidently occurs in other brain areas learning example in Scenario 1, in this old/new classifica-
[89,97,112,113]. tion problem we expect the steady state distribution of
synaptic weight values to be close to binary, that is,
This array of expectations and counter-expectations for consisting of only 2 categorically different long-term
synaptic plasticity mechanisms on Purkinje cell dendrites stable weight values (strong and weak) [21,153,154],
is collected in Table 1. though the absolute values of the weights might vary
according to the local dendritic input resistance
Scenario 2: A neuron that participates in an online, one-shot [12,17,18,21,128,129]. Finally, it was shown that a sub-
familiarity-based recognition memory task. stantial capacity boost can be realized in this type of
memory if LTD mechanisms preferentially target
those synapses that have been least recently potentiated
4
The Perceptron is a device that can be trained to decide whether an
input pattern (e.g. set of activated synapses) belongs to one class or
(since they carry the oldest stored information); this leads
another. It makes its predictions by combining a set of weights with the to the prediction that among the population of strong
input pattern through a dot product, and thresholding the result [105]. synapses, the time elapsed since a synapse’s most recent

Current Opinion in Neurobiology 2017, 43:177–186 www.sciencedirect.com


Synaptic plasticity in dendrites: complications and coping strategies Mel, Schiller and Poirazi 181

Figure 2

(a) Random connectivity Single layer "Perceptron"

x1
w1 weight plasticity
x2 w
2

Σ
in(t) out(t)
xk w
k

wo (t) = Q
xN w N

(b) Random connectivity 2-layer network with


conventional sigmoidal subunits

x1 w1
Σ weight plasticity
xk wk
.
.
Σ
out(t)
x1 w1
(t)
in(t)
Σ
xk wk

x1 w1
Σ
Strongly activated dendrite: xk wk
fires local spike
triggers local plasticity

(c) Structured connectivity 2-layer network with


spatially sensitive subunits

x1 w1
wiring plasticity
xk wk
.
.
Like-activated inputs x1 w1 Σ out(t)

form functional clusters in(t)

xk wk

x1 w1
Proximal-distal segregation
enables nonlinear xk wk
driver-modulator interactions

Current Opinion in Neurobiology

Schematics of 3 different learning scenarios.


(a) Scenario 1: a cell with random input connectivity is assumed to integrate its inputs linearly across the dendritic arbor, functioning as a single-
layer Perceptron. Image credit: NeuroMorpho.Org ID : NMO_10069

www.sciencedirect.com Current Opinion in Neurobiology 2017, 43:177–186


182 Neurobiology of learning and plasticity

potentiation event – which could be weeks or months – somatic action potentials (bAPs) to be instructive during
predicts its vulnerability to depression [118,120]. A can- early development (during the establishment of the clas-
didate biological mechanism is the ubiquitin-proteosome sical RF) but for local, dendritic potentials to regulate
system, which seems well suited to manage synapse aging plasticity later; and for LTP and LTD to act mainly as
through the elongation of ubiquitin chains, and has been arbiters of synapse survival or elimination during the
specifically suggested to orchestrate the dismantling of synapse spatial sorting process (Table 1).
synapses at the end of their lifetime [126,127,130–132].
Conclusions
Scenario 3: A neuron in the visual cortex that develops a Our starting point in this review has been that dendrites
‘complex’ receptive field (RF) with multiple RF subunits, each add tremendous complexity and diversity to neuronal
of which is modulated by subunit- specific contextual and/or plasticity mechanisms, relative to the simple scenario
attentional inputs. envisioned by Donald Hebb (Figure 1a). Our main thesis
has been that when it is possible to know or make
The receptive field (RF) of a “complex cell” (as defined reasonable guesses regarding the learning problem faced
by Hubel and Wiesel [133]) typically contains two or more by a particular neuron in a particular brain area, this
oriented ‘simple cell’ subunits, each modeled as a thre- knowledge allows us to make surprisingly strong infer-
sholded Gabor (or similar) filter [133–138]. Complex cells ences as to the role that dendrites may play in converting
exemplify the process whereby multiple similar ‘feature the cell’s inputs to outputs, and the way that plasticity
detectors’ (speaking loosely) are pooled to create spatial rules should operate within the cell’s dendritic tree to
invariance—a process that is repeated through the multi- establish the proper wiring diagram, or to set proper
ple stages of the ventral visual stream [139–141]. The synaptic weights, or both (Table 1, Figure 2). One of
possibility that the multiple subunits of a complex cell RF our main conclusions is that rather than expect stereo-
are mapped onto different dendrites has been discussed typed plasticity mechanisms, we should expect great
in the literature [97,137,142,143], and though it remains diversity in the way synaptic plasticity rules operate
speculative, we explore here the constraints that the idea within the dendrites of different neurons types within
imposes on the complex cell’s input–output function, as different brain areas.
well as the plasticity rules needed to set up the proper
wiring diagram (Figure 2c). One advantage of segregating
RF subunits onto different dendrites is that it permits Conflict of interest statement
each subunit to be targeted by distinct modulatory inputs. Nothing declared.
This is significant since it is known that both attentional
and contextual modulation of cells in early visual Acknowledgements
cortex can act at a scale finer than the overall RF, that This work was supported by the European Research Council Starting Grant
is, can modulate one RF subunit separately from the dEMORY (GA 311435) to P.P.; the Israeli Science Foundation, the Adelis
foundation and the Prince foundation (J.S.).
others [144–147]. Furthermore, both contextual and
attentional modulation effects are typically described
as ‘multiplicative’, in that they produce relatively little References and recommended reading
effect on their own, but increase the gain of the cell’s Papers of particular interest, published within the period of review,
have been highlighted as:
response to its classical driver inputs [144,148,149].
Subunit-specific gain boosting could be mediated by  of special interest
 of outstanding interest
nonlinear spatial summation effects in dendrites (see
Refs. [25,150] for a discussion of possible underlying 1. Hebb DO: The Organization of Behavior: A Neurophysiological
biophysical mechanisms). Establishing such a circuit dur- Theory. Wiley Press; 1949.
ing development would require a wiring plasticity rule 2. The Synaptic Organization of the Brain. Fifth Edition. Edited by
operating both (1) at the dendrite scale, to sort the inputs Gordon M Shepherd., The Quarterly Review of Biology 80, no. 4
(December 2005): 502–502. 10.1086/501315
corresponding to different RF subunits onto different
dendrites, and (2) at the subdendrite scale, to sort classical 3. Lund JS: Anatomical Organization of Macaque Monkey Striate
Visual Cortex. Annual Review of Neuroscience 1988, 11(1):253-
vs. modulatory inputs along the proximal distal axis of the 288 http://dx.doi.org/10.1146/annurev.ne.11.030188.001345.
dendrite. In this scenario, we expect the plasticity pro- 4. Andres-Barquin PJ, Ramón y Cajal: A century after the
cesses to operate mainly early in life; back propagating publication of his masterpiece. Endeavour 2001, 25:13-17.

(Figure 2 Legend Continued) (b) Scenario 2: a cell with random connectivity is shown with separately thresholded dendrites. In this case, the cell
acts as a conventional 2-layer neural network. When a dendrite receives strong excitation, a local spike is generated, and/or a local synaptic
plasticity event is triggered. Image credit: NeuroMorpho.Org ID : NMO_04127
(c) Scenario 3: a cell receives inputs that are spatially structured due to wiring plasticity. Structuring is schematized at two levels: (1) Correlated
inputs target the same dendrite (indicated by shared color scheme within each of three dendrites). (2) Within each dendrite, driver and modulator
inputs are shown spatially segregated. The dendritic nonlinearity (2-D mesh) in network shown at right represents within-dendrite nonlinear spatial
integration effects [150]. Image credit: Allen Institute

Current Opinion in Neurobiology 2017, 43:177–186 www.sciencedirect.com


Synaptic plasticity in dendrites: complications and coping strategies Mel, Schiller and Poirazi 183

5. Markram H, Toledo-Rodriguez M, Wang Y, Gupta A, Silberberg G, 23. Koch C, Poggio T, Torre V: Nonlinear interactions in a dendritic
Wu C: Interneurons of the neocortical inhibitory system. Nature tree: localization, timing, and role in information processing.
Reviews Neuroscience 2004, 5(10):793-807 http://dx.doi.org/ Proc. Natl. Acad. Sci. U. S. A. 1983, 80:2799-2802.
10.1038/nrn1519.
24. Gidon A, Segev I: Principles governing the operation of
6. Sandler M, Shulman Y, Schiller J: A novel form of local plasticity synaptic inhibition in dendrites. Neuron 2012, 75:330-341.
 in tuft dendrites of neocortical somatosensory layer
5 pyramidal neurons. Neuron 2016, 90:1028-1042. 25. M.P. Jadi, B.F. Behabadi, A. Poleg-Polsky, J. Schiller, B.W. Mel,
A unique plasticity mechanism for tuft dendrites of neocortical layer An Augmented Two-Layer Model Captures Nonlinear Analog
5 pyramidal neurons is described. The plasticity mechanism involves Spatial Integration Effects in Pyramidal Neuron Dendrites,
activation of NMDAR and Kv4.2 potassium channels, internalization of Proceedings of the IEEE. Institute of Electrical and Electronics
postsynaptic membrane and insertion of AMPAR to the postsynaptic Engineers, (2014) 102(5). 10.1109/JPROC.2014.2312671.
membrane. Importantly, this type of plasticity not only self-amplify the 26. Koch C, Poggio T, Torre V: Retinal ganglion cells: a functional
activated tuft synapses but also augment the local excitability of activated interpretation of dendritic morphology. Philos. Trans. R. Soc. L.
dendritic segments which in turn opens a prolonged plasticity window for B. Biol. Sci. 1982, 298:227-263.
additional convergent inputs.
27. Psarrou M, Stefanou SS, Papoutsi A, Tzilivaki A, Cutsuridis V,
7. Sajikumar S, Korte M: Different compartments of apical CA1 Poirazi P: A simulation study on the effects of dendritic
dendrites have different plasticity thresholds for expressing morphology on layer V prefrontal pyramidal cell firing
synaptic tagging and capture. Learn. Mem. 2011, 18:327-331. behavior. Front. Cell. Neurosci. 2014, 8.
8. Letzkus JJ, Kampa BM, Stuart GJ: Learning rules for spike 28. Johnston D, Magee JC, Colbert CM, Cristie BR, Christie BR:
timing-dependent plasticity depend on dendritic synapse Active properties of neuronal dendrites. Ann. Rev. Neurosci.
location. J. Neurosci. 2006, 26:10420-10429. 1996, 19:165-186.
9. Sjöström PJ, Häusser M: A cooperative switch determines the 29. Bernander O, Koch C, Douglas RJ: Amplification and
sign of synaptic plasticity in distal dendrites of neocortical linearization of distal synaptic input to cortical pyramidal cells.
pyramidal neurons. Neuron 2006, 51:227-238. J. Neurophysiol. 1994, 72:2743-2753.
10. Froemke RC, Poo M, Dan Y: Spike-timing-dependent synaptic 30. Cauller LJ, Connors BW: Functions of very distal dendrites:
plasticity depends on dendritic location. Nature 2005, 434:221- Experimental and computational studies of layer 1 synapses
225. on neocortical pyramidal cells. In Single Neuron Computation.
Edited by McKenna T, Davis J, Zornetzer SF. Academic Press;
11. Golding NL, Staff NP, Spruston N: Dendritic spikes as a 1992:199-229.
mechanism for cooperative long-term potentiation. Nature
2002, 418:326-331. 31. Polsky A, Mel BW, Schiller J: Computational subunits in thin
dendrites of pyramidal cells. Nat. Neurosci. 2004, 7:621-627.
12. Weber JP, Andrásfalvy BK, Polito M, Magó Á, Ujfalussy BB,
Makara JK: Location-dependent synaptic plasticity rules by 32. Harnett MT, Xu N-L, Magee JC, Williams SR: Potassium
dendritic spine cooperativity. Nat. Commun. 2016, 7:11380. Channels control the interaction between active dendritic
integration compartments in layer 5 cortical pyramidal
13. Gordon Polsky A, Schiller JU: Plasticity compartments in basal neurons. Neuron 2013, 79:516-529.
dendrites of neocortical pyramidal neurons. J. Neurosci. 2006,
26:12717-12726. 33. Larkum ME, Zhu JJ, Sakmann B: A new cellular mechanism for
coupling inputs arriving at different cortical layers. Nature
14. Rall W, Rinzel J: Branch input resistance and steady 1999, 398:338-341.
attenuation for input to one branch of a dendritic neuron
model. Biophys. J. 1973, 13:648-687. 34. Colbert CM, Magee JC, Hoffman DA, Johnston D: Slow recovery
from inactivation of Na+ channels underlies the activity-
15. Zador AM, Agmon-Snir H, Segev I: The morphoelectrotonic dependent attenuation of dendritic action potentials in
transform: a graphical approach to dendritic function. J. hippocampal CA1 pyramidal neurons. J. Neurosci. 1997,
Neurosci. 1995, 15:1669-1682. 17:6512-6521.

16. Araya R: Input transformation by dendritic spines of pyramidal 35. Frick A, Johnston D: Plasticity of dendritic excitability. J.
neurons. Front. Neuroanat. 2014, 8:141 http://dx.doi.org/ Neurobiol. 2005, 64:100-115.
10.3389/fnana.2014.00141.
36. Frick A, Magee J, Johnston D: LTP is accompanied by an
17. Branco Hausser MT: Synaptic integration gradients in single enhanced local excitability of pyramidal neuron dendrites. Nat.
cortical pyramidal cell dendrites. Neuron 2011, 69:885-892. Neurosci. 2004, 7:126-135.
37. Kim J, Hoffman DA: Potassium channels: newly found players in
18. Larkum Nevian T, Sandler M, Polsky A, Schiller JME: Synaptic
synaptic plasticity. Neuroscientist 2008, 14:276-286.
integration in tuft dendrites of layer 5 pyramidal neurons: a
new unifying principle. Science (80-) 2009, 325:756-760. 38. Kim Y, Hsu C-L, Cembrowski MS, Mensh BD, Spruston N:
Dendritic sodium spikes are required for long- term
19. Major G, Polsky A, Denk W, Schiller J, Tank DW: potentiation at distal synapses on hippocampal pyramidal
Spatiotemporally graded NMDA spike/plateau potentials in neurons. Elife 2015, 4.
basal dendrites of neocortical pyramidal neurons. J.
Neurophysiol. 2008, 99:2584-2601. 39. Losonczy A, Makara JK, Magee JC: Compartmentalized
dendritic plasticity and input feature storage in neurons.
20. Gulledge AT, Carnevale NT, Stuart GJ: Electrical advantages of Nature 2008, 452:436-441.
dendritic spines. PloS One 2012, 7(4):e36007 http://dx.doi.org/
10.1371/journal.pone.0036007. 40. Narayanan R, Johnston D: Functional maps within a single
neuron. J. Neurophysiol. 2012, 108:2343-2351.
21. Menon V, Musial TF, Liu A, Katz Y, Kath WL, Spruston N,
Nicholson DA: Balanced synaptic impact via distance- 41. Papoutsi A, Sidiropoulou K, Poirazi P: Memory beyond synaptic
dependent synapse distribution and complementary plasticity: the role of intrinsic neuronal excitability. Memory
expression of ampars and nmdars in hippocampal dendrites. Mechanisms in Health and Disease. World Scientific; 2012:: 53-80.
Neuron 2013, 80(6):1451-1463 http://dx.doi.org/10.1016/j.
neuron.2013.09.027. 42. Remy S, Beck H, Yaari Y: Plasticity of voltage-gated ion
channels in pyramidal cell dendrites. Curr. Opin. Neurobiol.
22. Magee JC, Cook EP: Somatic EPSP amplitude is independent 2010, 20:503-509.
of synapse location in hippocampal pyramidal neurons. Nature
Neuroscience 2000, 3(9):895-903 http://dx.doi.org/10.1038/ 43. Vaidya SP, Johnston D: Temporal synchrony and gamma-to-
78800. theta power conversion in the dendrites of CA1 pyramidal
neurons. Nat. Neurosci. 2013, 16:1812-1820.

www.sciencedirect.com Current Opinion in Neurobiology 2017, 43:177–186


184 Neurobiology of learning and plasticity

44. Kampa BM, Letzkus JJ, Stuart GJ: Dendritic mechanisms 60. Vu ET, Krasne FB: Evidence for a computational distinction
controlling spike-timing- dependent synaptic plasticity. between proximal and distal neuronal inhibition. Science 1992,
Trends Neurosci. 2007, 30:456-463. 255:1710-1712.
45. Song S, Miller KD, Abbott LF: Competitive Hebbian learning 61. Jadi M, Polsky A, Schiller J, Mel BW: Location-dependent
through spike- timing-dependent synaptic plasticity. Nat. effects of inhibition on local spiking in pyramidal neuron
Neurosci. 2000, 3:919-926. dendrites. PLoS Comput. Biol. 2012, 8:e1002550.

46. Lisman J, Spruston N: Postsynaptic depolarization 62. Konstantoudaki X, Papoutsi A, Chalkiadaki K, Poirazi P,
requirements for LTP and LTD: a critique of spike timing- Sidiropoulou K: Modulatory effects of inhibition on persistent
dependent plasticity. Nat. Neurosci. 2005, 8:839-841. activity in a cortical microcircuit model. Front. Neural Circuits
2014, 8:7.
47. Bi GQ, Poo MM: Synaptic modifications in cultured
hippocampal neurons: dependence on spike timing, synaptic 63. Ghiretti AE, Paradis S: Molecular mechanisms of activity-
strength, and postsynaptic cell type. J. Neurosci. 1998, dependent changes in dendritic morphology: role of RGK
18:10464-10472. proteins. Trends Neurosci. 2014, 37:399-407.

48. Vetter P, Roth A, Häusser M: Propagation of action potentials in 64. Harvey CD, Svoboda K: Locally dynamic synaptic learning rules
dendrites depends on dendritic morphology. J. Neurophysiol. in pyramidal neuron dendrites. Nature 2007, 450:1195-1200.
2001, 85:926-937.
65. Lyford GL, Yamagata K, Kaufmann WE, Barnes CA, Sanders LK,
49. Golding NL, Kath WL, Spruston N: Dichotomy of action-potential Copeland NG, Gilbert DJ, Jenkins NA, Lanahan AA, Worley PF:
backpropagation in CA1 pyramidal neuron dendrites. J. Arc, a growth factor and activity-regulated gene, encodes a
Neurophysiol. 2001, 86:2998-3010. novel cytoskeleton-associated protein that is enriched in
neuronal dendrites. Neuron 1995, 14:433-445.
50. Johnston D, Hoffman DA, Colbert CM, Magee JC: Regulation of
back-propagating action potentials in hippocampal neurons. 66. Steward O, Farris S, Pirbhoy PS, Darnell J, Van Driesche SJ:
Curr. Opin. Neurobiol. 1999, 9:288-292. Localization and local translation of Arc/Arg3.1 mRNA at
synapses: some observations and paradoxes. Front. Mol.
51. Takahashi H, Magee JC: Pathway interactions and synaptic Neurosci 2015:7.
plasticity in the dendritic tuft regions of CA1 pyramidal
neurons. Neuron 2009, 62(1):102-111 http://dx.doi.org/10.1016/j. 67. Pirbhoy PS, Farris S, Steward O: Synaptic activation of
neuron.2009.03.007. ribosomal protein S6 phosphorylation occurs locally in
activated dendritic domains. Learn. Mem. 2016, 23:255-269.
52. Sohal VS, Hasselmo ME: A model for experience-dependent
68. Govindarajan A, Israely I, Huang S-Y, Tonegawa S, Govindarajan
changes in the responses of inferotemporal neurons. Netw.
Israely I, Huang SY, Tonegawa SA: The dendritic branch is the
Comput. Neural Syst. 2000, 11:169-190.
preferred integrative unit for protein synthesis-dependent
53. Cichon J, Gan W-B: Branch-specific dendritic Ca(2+) spikes LTP. Neuron 2011, 69:132-146.
 cause persistent synaptic plasticity. Nature 2015, 520:180-185. 69. Sutton MA, Schuman EM: Local translational control in
This work highlights the importance of local dendritic NMDA-dependent dendrites and its role in long-term synaptic plasticity. J.
spikes in tuft dendrites of layer 5 pyramidal neurons along with specific Neurobiol. 2005, 64:116-131.
activation of somatostatin interneurons to motor learning tasks. Impor-
tantly, these branch specific NMDA-dependent spikes promote LTP in 70. Glazewski S, Giese KP, Silva A, Fox K: The role of alpha-CaMKII
active spines during the motor learning task. autophosphorylation in neocortical experience-dependent
plasticity. Nat. Neurosci. 2000, 3:911-918.
54. Sheffield MEJ, Dombeck DA: Calcium transient prevalence
 across the dendritic arbour predicts place field properties. 71. Makino H, Malinow R: Compartmentalized versus global
Nature 2015, 517:200-204. synaptic plasticity on dendrites controlled by experience.
This work provides direct dendritic, somatic and axonal calcium imaging Neuron 2011, 72:1001-1011.
from CA1 pyramidal neurons during place field formation in a head fixed
mice navigating in a virtual environment. Calcium transients indicative of 72. Shah MM: Cortical HCN channels: function, trafficking and
dendritic regenerative events were observed, and their prevalence was plasticity. J. Physiol. 2014, 592:2711-2719.
correlated with the spatial precision and stability of the neuronal place
fields. 73. Sachser RM, Crestani AP, Quillfeldt JA, Mello E, Souza T, de
Oliveira Alvares L: The cannabinoid system in the retrosplenial
55. Kawaguchi Y, Kubota Y: GABAergic cell subtypes and their cortex modulates fear memory consolidation,
synaptic connections in rat frontal cortex. Cereb. Cortex 1997, reconsolidation, and extinction. Learn. Mem. 2015, 22:584-588.
7:476-486.
74. Herring BE, Nicoll RA: Kalirin and Trio proteins serve critical
56. Yang GR, Murray JD, Wang X-J: A dendritic disinhibitory circuit roles in excitatory synaptic transmission and LTP. Proc. Natl.
mechanism for pathway-specific gating. Nat. Commun. 2016, Acad. Sci. U. S. A. 2016, 113:2264-2269.
7:12815.
75. Jensen CS, Watanabe S, Rasmussen HB, Schmitt N, Olesen S-P,
57. Kubota Y, Karube F, Nomura M, Kawaguchi Y: The diversity of Frost NA, Blanpied TA, Misonou H: Specific sorting and post-
cortical inhibitory synapses. Front. Neural Circuits 2016, 10:27. golgi trafficking of dendritic potassium channels in living
neurons. J. Biol. Chem. 2014, 289:10566-10581.
58. Bloss EB, Cembrowski MS, Karsh B, Colonell J, Fetter RD,
 Spruston N: Structured dendritic inhibition supports branch- 76. Mel BW: The clusteron: toward a simple abstraction for a
selective integration in CA1 pyramidal cells. Neuron 2016, complex neuron. In Advances in Neural Information Processing
89:1016-1030. Systems. Edited by Moody J, Hanson S, Lippmann R, Morgan
This work shows that synaptic connectivity between molecularly defined umn. 1992:35-42.
inhibitory interneurons and CA1 pyramidal cell dendrites can be highly
77. Poirazi P, Mel BW: Impact of active dendrites and structural
selective. Interneurons can specifically target dendritic branch types, as
plasticity on the memory capacity of neural tissue. Neuron
well as specific locations like the origin or end of individual branches.
2001, 29:779-796.
Computational simulations indicate that the observed subcellular target-
ing enables control over the nonlinear integration of synaptic input or the 78. Stepanyants A, Hof PR, Chklovskii DB: Geometry and structural
initiation and backpropagation of action potentials in a branch-selective plasticity of synaptic connectivity. Neuron 2002, 34:275-288.
manner. Thus, interneurons may play a key role in shaping dendritic
computations carried out by principal cells. 79. Chklovskii DB, Mel BW, Svoboda K: Cortical rewiring and
information storage. Nature 2004, 431:782-788.
59. DeFelipe J, Ballesteros-Yáñez I, Inda MC, Muñoz A: Double-
bouquet cells in the monkey and human cerebral cortex with 80. Legenstein R, Maass W: Branch-specific plasticity enables self-
special reference to areas 17 and 18. Prog. Brain Res. 2006, organization of nonlinear computation in single neurons. J.
154:15-32. Neurosci. 2011, 31:10787-10802.

Current Opinion in Neurobiology 2017, 43:177–186 www.sciencedirect.com


Synaptic plasticity in dendrites: complications and coping strategies Mel, Schiller and Poirazi 185

81. Kastellakis G, Silva AJ, Poirazi P: Linking memories across time 99. Smith SL, Smith IT, Branco T, Häusser M: Dendritic spikes
 via neuronal and dendritic overlaps in model neurons with enhance stimulus selectivity in cortical neurons in vivo. Nature
active dendrites. Cell Rep. 2016, 17:1491-1504. 2013, 503(7474):115-120 http://dx.doi.org/10.1038/nature12600.
This work describes a large-scale, biologically constrained, network
model of an associative brain area that incorporates non-linear dendrites 100. Lee D, Lin B-J, Lee AK: Hippocampal place fields emerge upon
and multiple plasticity rules. The model is used to explore the rules single-cell manipulation of excitability during behavior.
underlying the linking of associative memories across time. It is predicted Science 2012, 337:849-853.
that memories become linked via their storage in overlapping neuronal
populations, whereby they form synaptic clusters. Synapse clustering is 101. Remy S, Spruston N: Dendritic spikes induce single-burst long-
proposed to underlie memory linking at the levels of a single, a weak and a term potentiation. Proc. Natl. Acad. Sci. U. S. A. 2007,
strong as well as multiple memories, serving as a mechanism for the 104:17192-17197.
formation of episodic memories. 102. Albus JS: A theory of cerebellar function. Math. Biosci. 1971,
82. González J, Mel B, Poirazi P: Distinguishing linear vs: non-linear 10:25-61.
integration in CA1 radial oblique dendrites: it’s about time. 103. Marr D: A theory of cerebellar cortex. J. Physiol. 1969, 202:437-
Front. Comput. Neurosci. 2011, 5:1-12. 470.
83. Iannella NL, Launey T, Tanaka S: Spike timing-dependent 104. Kanerva P: Sparse Distributed Memory. The MIT Press; 2003.
plasticity as the origin of the formation of clustered synaptic
efficacy engrams. Front Comput Neurosci. 2010, 4 pii: 21. 105. Rosenblatt F: The perceptron: a probabilistic model for
information storage and organization in the brain. Cornell
84. Wiegert JS, Oertner TG: Long-term depression triggers the Aeronaut. Lab. Psychol. Rev. 1958, 65:386-408.
selective elimination of weakly integrated synapses. Proc. Nat.
Acad. Sci. 2013, 110(47):E4510-E4519 http://dx.doi.org/10.1073/ 106. Tyrrell T, Willshaw D: Cerebellar cortex: its simulation and the
pnas.1315926110. relevance of Marr’s theory. Philos. Trans. R. Soc. B Biol. Sci.
1992, 336:239-257.
85. Oh WC, Parajuli LK, Zito K: Heterosynaptic Structural Plasticity
on Local Dendritic Segments of Hippocampal C Neurons. Cell 107. Clopath C, Brunel N: Optimal properties of analog perceptrons
Reports 2015, 10(2):162-169 http://dx.doi.org/10.1016/j. with excitatory weights. PLoS Comput. Biol. 2013, 9:e1002919.
celrep.2014.12.016.
108. Cover TM: Geometrical and statistical properties of systems of
86. Larkum ME, Nevian T: Synaptic clustering by dendritic linear inequalities with applications in pattern recognition.
signalling mechanisms. Curr. Opin. Neurobiol. 2008, 18:321-331. IEEE Trans. Electron. Comput. 1965, 14:326-334.
87. Kleindienst T, Winnubst J, Roth-Alpermann C, Bonhoeffer T, 109. Vapnik VN, Chervonenkis AY: On the uniform convergence of
Lohmann C: Activity-dependent clustering of functional relative frequencies of events to their probabilities. Theory
synaptic inputs on developing hippocampal dendrites. Neuron Probab. Appl. 1971, 16:264-280.
2011, 72(6):1012-1024 http://dx.doi.org/10.1016/j.
neuron.2011.10.015. 110. Titley HK, Hansel C: Asymmetries in cerebellar plasticity and
motor learning. Cerebellum 2016, 15:87-92.
88. Kastellakis G, Cai DJ, Mednick SC, Silva AJ, Poirazi P: Synaptic
clustering within dendrites: an emerging theory of memory 111. Brunel N, Hakim V, Isope P, Nadal J-P, Barbour B: Optimal
formation. Prog. Neurobiol. 2015, 126:19-35. information storage and the distribution of synaptic weights.
Neuron 2004, 43:745-757.
89. DeBello WM, McBride TJ, Nichols GS, Pannoni KE, Sanculi D,
Totten DJ: Input clustering and the microscale structure of 112. Fu M, Yu X, Lu J, Zuo Y: Repetitive motor learning induces
local circuits. Front. Neural Circuits 2014, 8:112. coordinated formation of clustered dendritic spines in vivo.
Nature 2012, 483:92-95.
90. Schiller J, Major G, Koester HJ, Schiller Y: NMDA spikes in basal
dendrites of cortical pyramidal neurons. Nature 2000, 404:285- 113. Winnubst J, Lohmann C: Synaptic clustering during
289. development and learning: the why, when, and how. Front. Mol.
Neurosci. 2012, 5.
91. Antic SD, Zhou W-L, Moore AR, Short SM, Ikonomu KD: The
decade of the dendritic NMDA spike. J. Neurosci. Res. 2010, 114. Amit DJ, Fusi S: Learning in neural networks with material
88:2991-3001. synapses. Neural Comput. 1994, 6:957-982.
92. Mel BW: Synaptic integration in an excitable dendritic tree. J. 115. Benna MK, Fusi S: Computational principles of synaptic
Neurophysiol. 1993, 70:1086-1101. memory consolidation. Nat. Neurosci. 2016, 19(12):1697-1706
http://dx.doi.org/10.1038/nn.4401.
93. Mel BW: NMDA-Based pattern discrimination in a modeled
cortical neuron. Neural Comput. 1992, 4:502-516. 116. Fusi S, Abbott LF: Limits on the memory storage capacity of
bounded synapses. Nature Neuroscience 2007, 10(4):485-493.
94. Poirazi P, Brannon T, Mel BW: Pyramidal neuron as two-layer
neural network. Neuron 2003, 37:989-999. 117. Fusi S, Drew PJ, Abbott LF: Cascade models of synaptically
stored memories. Neuron 2005, 45:599-611.
95. Poirazi P, Brannon T, Mel BW: Arithmetic of subthreshold
synaptic summation in a model CA1 pyramidal cell. Neuron 118. Henson RN, Willshaw DJ: Short-term associative memory.
2003, 37:977-987. Proceedings of the INNS World Congress on Neural Networks
1995:438-441.
96. Lavzin M, Rapoport S, Polsky A, Garion L, Schiller J: Nonlinear
dendritic processing determines angular tuning of barrel 119. Nadal JP, Toulouse G, Changeux JP, Dehaene S: Networks of
cortex neurons in vivo. Nature 2012, 490:397-401. formal neurons and memory palimpsests. EPL (Europhys. Lett.)
1986, 1:535.
97. Wilson DE, Whitney DE, Scholl B, Fitzpatrick D: Orientation
selectivity and the functional clustering of synaptic inputs in 120. Wu XE, Mel BW: Capacity-enhancing synaptic learning rules in
primary visual cortex. Nat. Neurosci. 2016, 19:1003-1009. a medial temporal lobe online learning model. Neuron 2009,
62:31-41.
98. Takahashi N, Oertner TG, Hegemann P, Larkum ME: Active
 cortical dendrites modulate perception. Science (80-) 2016, 121. Brown MW, Aggleton JP: Recognition memory: what are the
354:1587-1590. roles of the perirhinal cortex and hippocampus? Nat. Rev.
This work is the first to show that Ca2+ activity in the apical dendrites of a Neurosci. 2001, 2:51-61.
subset of layer 5 (L5) pyramidal neurons in primary somatosensory cortex
(S1) in mice is correlated with the threshold for perceptual detection of 122. Eichenbaum H, Sauvage M, Fortin N, Komorowski R, Lipton P:
whisker deflections. Manipulating the activity of apical dendrites shifts the Towards a functional organization of episodic memory in the
perceptual threshold, demonstrating that an active dendritic mechanism medial temporal lobe. Neurosci. Biobehav. Rev. 2012,
is causally linked to perceptual detection. 36:1597-1608.

www.sciencedirect.com Current Opinion in Neurobiology 2017, 43:177–186


186 Neurobiology of learning and plasticity

123. Mel GC, Mel BW: Comparing memory models with and without 140. Riesenhuber M, Poggio T: Hierarchical models of object
dendrites. Society for Neuroscience Abstract #838.16. CA: San recognition in cortex. Nat. Neurosci. 1999, 2:1019-1025.
Diego; 2016.
141. Rust NC, DiCarlo JJ: Selectivity and tolerance (“invariance”)
124. Frey U, Morris RG: Synaptic tagging and long-term both increase as visual information propagates from cortical
potentiation. Nature 1997, 385:533-536. area V4 to IT. J. Neurosci. 2010, 30:12978-12995.
125. Govindarajan A, Kelleher RJ, Tonegawa S: A clustered plasticity 142. Hubel DH: Exploration of the primary visual cortex, 1955–78.
model of long-term memory engrams. Nat. Rev. Neurosci. 2006, Nature 1982, 299:515-524.
7:575-583.
143. Mel BW, Ruderman DL, Archie KA: Translation-invariant
126. Sachser RM, Santana F, Crestani AP, Lunardi P, Pedraza LK, orientation tuning in visual “complex” cells could derive from
Quillfeldt JA, Hardt O, de Oliveira Alvares L: Forgetting of long- intradendritic computations. J. Neurosci. 1998, 18:4325-4334.
term memory requires activation of NMDA receptors L-type
voltage-dependent Ca2+ channels, and calcineurin. Sci. Rep. 144. Kapadia MK, Ito M, Gilbert CD, Westheimer G: Improvement in
2016, 6:22771. visual sensitivity by changes in local context: parallel studies
in human observers and in V1 of alert monkeys. Neuron 1995,
127. Migues PV, Liu L, Archbold GEB, Einarsson EO, Wong J,
15:843-856.
Bonasia K, Ko SH, Wang YT, Hardt O: Blocking synaptic removal
of GluA2-containing AMPA receptors prevents the natural 145. Moran J, Desimone R: Selective attention gates visual
forgetting of long-term memories. J. Neurosci. 2016, processing in the extrastriate cortex. Science 1985,
36:3481-3494. 229:782-784.
128. Katz Y, Menon V, Nicholson DA, Geinisman Y, Kath WL,
146. Treue S, Maunsell JH: Attentional modulation of visual motion
Spruston N: Synapse distribution suggests a two-stage model
processing in cortical areas MT and MST. Nature 1996,
of dendritic integration in CA1 pyramidal neurons. Neuron
382:539-541.
2009, 63:171-177.
129. Nevian T, Larkum ME, Polsky A, Schiller J: Properties of basal 147. Reynolds JH, Chelazzi L, Desimone R: Competitive mechanisms
dendrites of layer 5 pyramidal neurons: a direct patch-clamp subserve attention in macaque areas V2 and V4. J. Neurosci.
recording study. Nat. Neurosci. 2007, 10:206-214. 1999, 19:1736-1753.

130. Colledge M, Snyder EM, Crozier RA, Soderling JA, Jin Y, 148. McAdams CJ, Maunsell JH: Effects of attention on orientation-
Langeberg LK, Lu H, Bear MF, Scott JD: Ubiquitination regulates tuning functions of single neurons in macaque cortical area
PSD-95 degradation and AMPA receptor surface expression. V4. J. Neurosci. 1999, 19:431-441.
Neuron 2003, 40:595-607.
149. Reynolds JH, Heeger DJ: The normalization model of attention.
131. Ding M, Shen K: The role of the ubiquitin proteasome system in Neuron 2009, 61:168-185.
synapse remodeling and neurodegenerative diseases.
Bioessays 2008, 30:1075-1083. 150. Jadi M, Behabadi BF, Poleg-Polsky A, Schiller J, Mel BW: An
augmented two-layer model captures nonlinear analog spatial
132. Tai H-C, Schuman EM: Ubiquitin, the proteasome and protein integration effects in pyramidal neuron dendrites. Proc. IEEE
degradation in neuronal function and dysfunction. Nat. Rev. 2014, 102:782-798.
Neurosci. 2008, 9:826-838.
151. Jelitai M, Puggioni P, Ishikawa T, Rinaldi A, Duguid I: Dendritic
133. Hubel DH, Wiesel TN: Receptive fields, binocular interaction excitation–inhibition balance shapes cerebellar output during
and functional architecture in the cat’s visual cortex. J. Physiol. motor behaviour. Nat. Comm. 2016, 7:13722 http://dx.doi.org/
1962, 148:574-591. 10.1038/ncomms13722.
134. Palmer LA, Davis TL: Receptive-field structure in cat striate 152. Walter JT, Khodakhah K: The linear computational algorithm of
cortex. J. Neurophysiol. 1981, 46:260-276. cerebellar purkinje cells. J. Ceurosc. 2006, 26(50):12861-12872
http://dx.doi.org/10.1523/JNEUROSCI.4507-05.2006.
135. Adelson EH, Bergen JR: Spatiotemporal energy models for the
perception of motion. J. Opt. Soc. Am. A 1985, 2:284. 153. O’Connor DH, Wittenberg GM, Wang SS-H: Graded bidirectional
136. Ohzawa I, Sclar G, Freeman RD: Contrast gain control in the synaptic plasticity is composed of switch-like unitary events.
cat’s visual system. J. Neurophysiol. 1985, 54:651-667. Proc. Nat. Acad. Sci. 2005, 102(27):9679-9684 http://dx.doi.org/
10.1073/pnas.0502332102.
137. Archie KA, Mel BW: A model for intradendritic computation of
binocular disparity. Nat. Neurosci. 2000, 3:54-63. 154. Petersen CCH, Malenka RC, Nicoll RA, Hopfield JJ: All-or-none
potentiation at CA3-CA1 synapses. Sci. 1998, 95(8):4732-4737.
138. Fregnac Y, Shulz DE: Activity-dependent regulation of
receptive field properties of cat area 17 by supervised Hebbian 155. Sjöström PJ, Rancz EA, Roth A, Häusser M: Dendritic excitability
learning. J. Neurobiol. 1999, 41:69-82. and synaptic plasticity. Physiol. Rev. 2008, 88(2):769-840 http://
dx.doi.org/10.1152/physrev.00016.2007 Review. PMID:
139. Fukushima K: Neocognitron: a self organizing neural network 18391179.
model for a mechanism of pattern recognition unaffected by
shift in position. Biol. Cybern. 1980, 36:193-202.

Current Opinion in Neurobiology 2017, 43:177–186 www.sciencedirect.com

You might also like