Upload
trinhphuc
View
215
Download
1
Embed Size (px)
Citation preview
UNIVERSIDADE FEDERAL DO RIO GRANDE DO NORTE
PROGRAMA DE PÓS-GRADUAÇÃO EM NEUROCIÊNCIAS
I N S T I T U TO D O !C É R E B RO
ENCODING MECHANISMS BASED ON FAST OSCILLATIONS IN THE RETINA OF THE CAT AND THEIR DEPENDENCIES ON
ANESTHESIA
F A B I O F R E I T A G
N a t a l , 2 0 1 3
UNIVERSIDADE FEDERAL DO RIO GRANDE DO NORTE
PROGRAMA DE PÓS-GRADUAÇÃO EM NEUROCIÊNCIAS
I N S T I T U TO D O !C É R E B RO
ENCODING MECHANISMS BASED ON FAST OSCILLATIONS IN THE RETINA OF THE CAT AND THEIR DEPENDENCIES ON
ANESTHESIA
F A B I O F R E I T A G
TRABALHO APRESENTADO AO PROGRAMA DE PÓS-GRADUAÇÃO EM NEUROCIÊNCIAS DA UNIVERSIDADE
FEDERAL DO RIO GRANDE DO NORTE COMO REQUISITO PARCIAL PARA A OBTENÇÃO DO
G R A U D E M E S T R E
ORIENTADOR: Prof. Dr. SERGIO NEUENSCHWANDER
NEUROBIOLOGIA DE SISTEMAS E COGNIÇÃO
N a t a l , 2 0 1 3
UNIVERSIDADE FEDERAL DO RIO GRANDE DO NORTE
PROGRAMA DE PÓS-GRADUAÇÃO EM NEUROCIÊNCIAS
I N S T I T U TO D O !C É R E B RO
ENCODING MECHANISMS BASED ON FAST OSCILLATIONS IN THE RETINA OF THE CAT AND THEIR DEPENDENCIES ON
ANESTHESIA
F A B I O F R E I T A G
DISSERTAÇÃO APRESENTADA AO PROGRAMA DE PÓS-GRADUAÇÃO EM NEUROCIÊNCIAS
DA UNIVERSIDADE FEDERAL DO RIO GRANDE DO NORTE
COMO REQUISITO PARCIAL PARA A OBTENÇÃO DO GRAU DE MESTRE. ÁREA DE CONCENTRAÇÃO: NEUROBIOLOGIA DE SISTEMAS E COGNIÇÃO.
APROVADA EM: 27.08.2013
B A N C A E X A M I N A D O R A : PROF.DR. SERGIO NEUENSCHWANDER
PROF. DR. ADRIANO TORT
PROF. DR. JEROME BARON
AB S T R A C T
Processing in the visual system starts in the retina. Its complex network of cells with different properties enables for parallel encoding and transmission of visual information to the lateral geniculate nucleus (LGN) and to the cortex. In the retina, it has been shown that responses are often accompanied by fast synchronous oscillations (30 - 90 Hz) in a stimulus-dependent manner. Studies in the frog, rabbit, cat and monkey, have shown strong oscillatory responses to large stimuli which probably encode global stimulus properties, such as size and continuity (Neuenschwander and Singer, 1996; Ishikane et al., 2005). Moreover, simultaneous recordings from different levels in the visual system have demonstrated that the oscillatory patterning of retinal ganglion cell responses are transmitted to the cortex via the LGN (Castelo-Branco et al., 1998). Overall these results suggest that feedforward synchronous oscillations contribute to visual encoding.
In the present study on the LGN of the anesthetized cat, we further investigate the role of retinal oscillations in visual processing by applying complex stimuli, such as natural visual scenes, light spots of varying size and contrast, and flickering checkerboards. This is a necessary step for understanding encoding mechanisms in more naturalistic conditions, as currently most data on retinal oscillations have been limited to simple, flashed and stationary stimuli. Correlation analysis of spiking responses confirmed previous results showing that oscillatory responses in the retina (observed here from the LGN responses) largely depend on the size and stationarity of the stimulus. For natural scenes (gray-level and binary movies) oscillations appeared only for brief moments probably when receptive fields were dominated by large continuous, flat-contrast surfaces. Moreover, oscillatory responses to a circle stimulus could be broken with an annular mask indicating that synchronization arises from relatively local interactions among populations of activated cells in the retina.
A surprising finding in this study was that retinal oscillations are highly dependent on halothane anesthesia levels. In the absence of halothane, oscillatory activity vanished independent of the characteristics of the stimuli. The same results were obtained for isoflurane, which has similar pharmacological properties. These new and unexpected findings question whether feedfoward oscillations in the early visual system are simply due to an imbalance between excitation and inhibition in the retinal networks generated by the halogenated anesthetics. Further studies in awake behaving animals are necessary to extend these conclusions.
Key words: retina, oscillation, synchronization, halothane, visual.
RESUMO
O processamento da informação visual se inicia na retina. A sua complexa rede de células com diferentes propriedades permite que a informação visual seja codificada em canais paralelos e transmitida para o núcleo geniculado lateral (LGN) e o córtex. Na retina, tais respostas estão frequentemente acompanhadas por oscilações sincronizadas de alta frequência (30 – 90 Hz) em uma maneira dependente do estímulo. Como demonstrado em estudos na rã, coelho, gato e macaco, respostas oscilatórias ocorrem em geral a estímulos relativamente grandes, podendo codificar propriedades globais do estímulo como o tamanho e continuidade (Neuenschwander and Singer, 1996; Ishikane et al., 2005). Além disso, registros simultâneos em diferentes níveis do sistema visual têm mostrado que o padrão de oscilação nas células ganglionares retinianas é transmitido para o córtex visual via LGN (Castelo-Branco et al., 1998). De uma forma geral, esses resultados sugerem que oscilações sincronizadas em uma maneira feedforward são importantes na codificação da informação visual.
No presente estudo feito no LGN de gatos anestesiados, investigamos o papel das oscilações retinianas no processamento de informação visual através da apresentação de estímulos complexos, como cenas naturais, pixels aleatórios no tempo e espaço, além de grades em movimento. Esse é um importante passo para o entendimento de mecanismos de codificação em condições naturais, já que grande parte dos estudos que investigaram o papel de oscilações retinianas utilizaram-se de estímulos simples e estacionários. Análises de correlação de respostas neuronais (spiking responses) confirmaram resultados prévios mostrando que respostas oscilatórias na retina (observadas aqui a partir de registros no LGN) dependem do tamanho e estacionariedade do estímulo. Para filmes de cenas naturais (em escala de cinza e preto e branco) oscilações apareceram apenas por breves momentos provavelmente quando os campos receptores foram dominados por padrões extensos e contínuos (para ambas as escalas). As atividades oscilatórias parecem ser dependentes de uma massa crítica de células ativadas sugerindo que esse padrão regular de atividade surge através de interações horizontais na retina.
Nossos resultados mostram, além disto, que surpreendentemente oscilações da retina no gato são dependentes da anestesia mediada por halotano. Na ausência deste, atividades oscilatórias estiveram ausentes independentemente das características dos estímulos visuais. Resultados semelhantes foram obtidos para o isoflurano, anestésico com propriedades farmacológicas similares. Esse novo e inesperado resultado nos faz questionar se oscilações feedforward no sistema visual não seriam resultado de um desequilíbrio entre correntes de excitação e inibição nas redes retinianas gerado pelos anestésicos halogenados. Experimentos futuros em animais acordados serão necessários para confirmar essas conclusões.
Palavras-chave: retina, oscilação, sincronização, halotana, visão.
LI S T O F F I G U R E S A N D TA B L E S
Figure 1│ Schematic representation of the LGN in the cat.
Figure 2│ Stimulus size modulates spiking responses of an ON-center cell in the LGN.
Figure 3│ The binding by synchronization hypothesis.
Figure 4│ Oscillatory responses in the retina.
Figure 5│ Synchronous oscillations in the retina arises from population interactions.
Figure 6│ Feedforward synchronization in the retinogeniculate pathway.
Figure 7│ Synchronous oscillations in the retina encode stimulus continuity
Figure 8│ Schematic representation of the electrode recording system used in this study.
Figure 9│ Quartz-insulated tungsten-platinum electrodes.
Figure 10│ Stimulus size and circle annular mask protocols.
Figure 11│ Percolation stimulus protocol.
Figure 12│ Dynamical visual stimuli.
Figure 13│ Stimulus size and luminance modulate fast oscillations in the retinogeniculate system.
Figure 14│ Correlated activity across large distances in the retina.
Figure 15│ Breaking stimulus continuity disrupts oscillatory responses to a circle stimulus.
Figure 16│ Percolation threshold does not lead to a phase transition in the oscillatory responses.
Figure 17│ Fast oscillations appear only briefly for homogeneous image segments in natural scene movies.
Figure 18│ Oscillatory responses to gratings as function of stimulus temporal frequency.
Figure 19│ Halothane anesthesia greatly affects fast oscillations in the retina.
Figure 20│ Effects of halothane on LGN responses to a linear ramp stimulus.
Figure 21│ Oscillation strength as a function of halothane levels for responses to a linear ramp stimulus. Population data.
Figure 22│ Retinal oscillations in response to natural scene movies vanish in absence of halothane.
Figure 23│ Oscillation strength as a function of halothane levels for responses to natural scene movies. Population data.
Figure 24│ Retinal oscillations in LGN responses to grating stimuli.
Figure 25│ Oscillation strength as a function of halothane levels for responses to grating stimuli. Population data.
Figure 26│ Isoflurane, an halogenated anesthetic similar to halothane, is also responsible for the appearance of fast oscillations in the retinogeniculate system.
Figure 27│ Example of a small-world network.
Figure 28│ Visual processing using coupled oscillators.
Table 1│ Mean values and standard deviation for modulation amplitude and firing rate of the 10 different sizes in 3 different levels of luminance.
Table 2│ Means and standard deviation for modulation amplitude and firing rate for masked and unmasked circular stimulus.
SUMMARY
AB S T R A C T A N D K E Y W O R D S …………..…….…….………………… 04
RE S U M O E PA L AV R A S-C H AV E …………..…….…….……………… 05
LI S T O F F I G U R E S A N D TA B L E S ………….….….….………………… 06
SU M M A RY ……………………..…….….….….…………………… 07
IN T R O D U C T I O N ……………………..…….….….….……………… 09
1.1 PA R A L L E L C H A N N E L S I N T H E R E T I N O G E N I C U L AT E S Y S T E M ………… 10
1.2 NE U R O N A L E N C O D I N G …………………………….……….……… 11
1.3 TH E B I N D I N G B Y S Y N C H R O N I Z AT I O N H Y P O T H E S I S …………..…........ 13
1.4 SY N C H R O N O U S O S C I L L AT I O N S I N T H E E A R LY V I S U A L S Y S T E M .............. 15
1.5 RE T I N A L O S C I L L AT I O N S A S A N E N C O D I N G M E C H A N I S M ........................ 18
1.6 AN E S T H E S I A E F F E C T S…………………….………….………….… 20
2. OB J E C T I V E S ……………………..…….….….….……………… 21
Specif ic goals ……….……….……………….……….…………… 21
3. ME T H O D S ………………………………………………………… 22
3.1 EX P E R I M E N TA L S E S S I O N S…………..…..…..…..…..…..…..…..…. 22
3.1 .1 SU R G I C A L P R O C E D U R E S , A N E S T H E S I A A N D L I F E S U P P O RT… 22
3.1 .2 RE C O R D I N G D E V I C E A N D E L E C T R O D E S .….…....…..…..…. 23
3.1 .3 S I G N A L C O N D I T I O N I N G A N D D ATA A C Q U I S I T I O N…………… 25
3.2 . V I S U A L S T I M U L I ………………………………………………… 26
Circle size …………………………………………………… 27
Circle annular mask ………………………………………… 27
Percolation stimulus ………………………………………… 27
Natural movies and gratings ………………………………… 28
3.3 . HA L O T H A N E E F F E C T S ……………………………………………… 29
3.4 . DATA A N A LY S I S …………………………………………………… 29
4. RE S U LT S …………………………………………………………… 31
4.1 . DE P E N D E N C I E S O N S T I M U L U S S I Z E ………………………………… 31
4.2 . BR E A K I N G S T I M U L U S C O N T I N U I T Y ………………………………… 33
4.3 . PE R C O L AT I O N S T I M U L U S ………………………………………… 34
4.4 . OS C I L L AT O RY R E S P O N S E S T O DY N A M I C A L S T I M U L I ………………… 35
4.5 . DE P E N D E N C I E S O N H A L O T H A N E L E V E L S………… ………………… 37
5. D I S C U S S I O N ………… …………………………………………… 43
5.1 ST I M U L U S D E P E N D E N C I E S…………………………………………… 44
5.2 EF F E C T S O F H A L O T H A N E A N E S T H E S I A………………………………… 47
6. CO N C L U S I O N ………………………………………………………… 49
7. RE F E R E N C E S ………………...………………………………………… 50
8. TA B L E S……………………...………………………………………… 56
9. AC K N O W L E D G M E N T S……….………………………………………… 58
1. IN T R O D U C T I O N
It is impossible to think about how we interact with the world without focusing
on sensory systems. Our personal sensory experiences and the memories resulting from
them contribute directly to shaping our identity. Moreover, perceived environmental
cues like odors, colors, shapes and sounds, are essential for building appropriate
responses during our social interactions. Being able to recognize a threatening predator,
a potential prey, or a receptive female, for example, are crucial for the maintenance of
the life of an individual and eventually transmission of its genes to the next generations.
Given their central importance, its is not surprising to see that during evolution different
sensory modalities appeared associated with distinct neural machinery. Yet, despite
being represented by multiple inputs, objects are perceived not as a collection of
independent attributes but as unified percepts. What are the neural mechanisms
responsible for the representation and coordination of inputs into coherent percepts?
One strategy that the brain adopted to represent sensory information was through
parallel processing (Nassi and Callaway, 2009). This notion was first put forward for the
somatosensory system, where the sensations of pain and temperature were found to be
transmitted independently through parallel pathways (Gasser and Erlanger, 1929). This
concept was later extended to the visual system based on the observation that at least
three groups of axons (with distinct calibre and conduction velocities) could be
identified in the optic nerve of the rabbit (Bishop, 1933). The concept of parallel
pathways was further supported by studies in the visual system of cats and monkeys
showing independent forward processing channels from the retina to the lateral
geniculate nucleus (LGN) and to the primary visual cortex (Casagrande and Kaas, 1994;
Sherman and Guillery, 2006). In addition to parallel channels, the visual system is
organized in multiple areas, which provide multiple representations of the visual field
(Gattass et al., 2011).
Despite all the data accumulated on how neural systems represent environmental
bits of information we still know very little about the nature of the integrative process in
the brain. In particular we do not know how perceptual binding is achieved based on the
highly distributed architecture of the sensory systems.
Fast osci l lat ions in the ret ina by Fabio Freitag 9
1.1 PA R A L L E L C H A N N E L S I N T H E R E T I N O G E N I C U L AT E S Y S T E M
Even a simple visual scene can be decomposed into distinct attributes such as
brightness, color, depth, shape and motion. Processing all this information in parallel
implies into representing specific features by appropriate channels, from the first
processing steps at the level of the retina to the primary visual cortex and extrastriate
areas (Livingstone and Hubel, 1988).
In the retina, evidence for parallel processing was first obtained by Christina
Enroth-Cugell and John Robson with their seminal work in the cat (Enroth-Cugell and
Robson, 1966). In response to light, retinal photoreceptors change their membrane
potential. These signals are then transmitted through the bipolar cells to the ganglion
cells, where action potentials are generated and sent to the retinogeniculate pathway.
Distinct pathways are originated from the different types of retinal ganglion cells, which
convey parallel information first to the lateral geniculate nucleus (LGN) and then to the
visual cortex (Yeh et al., 2003). In the cat, the two most important functional cell types
are the X- and Y-cells. These cell types differ primarily in their morphology and
response properties. X-cells show sustained responses to spot stimuli and have dendritic
arbors confined to a narrow area, while Y-cells respond transiently and have wide
receptive fields (RFs) (Saito, 2004). Furthermore, X-cells exhibit a linear summation of
responses for stimuli presented to different subregions of the RF. In contrast, Y-cells do
not integrate linearly the spatial information (Kratz et al., 1978).
In the cat LGN, principal cells were also classified as X- or Y-cells (Hoffmann et
al., 1972). As in the retina, X-cells are less sensitive to fast movements of the stimulus,
while Y-cells show strong modulation to motion since they are capable of responding to
stimuli of high temporal frequency. Moreover, X-cells are capable of responding to
stimuli of high spatial frequencies, while Y-cells respond vigorously to stimuli of low
spatial frequency. Thus, we can infer that the Y-channel is important to encode visual
aspects related to motion processing, while X-cells are related to processing information
such as form (Tolhurst, 1973; Stone, 1983).
The two functional classes of cells in retina also differ in their projections to the
LGN. X-cells project mainly to the contralateral layer A and Y-cells diverge to project to
layers A and C, both contralateral (Yeh et al., 2003). Commonly known as a first-order
Fast osci l lat ions in the ret ina by Fabio Freitag 10
relay, the LGN is a laminated structure localized in the thalamus, conveying visual
information from the retina to the cortex. In the cat, the LGN is composed of three main
layers: A, A1 and C, exhibiting a retinotopic organization with emphasis in the
representation of the the central region of the visual field (Sherman and Guillery, 2002,
see Figure 1). The receptive field (RF) structure of LGN neurons is concentric, with an
antagonistic center-surround organization, similar to those observed in retinal ganglion
cells, and with either ON-center or OFF-center (Hubel and Wiesel, 1962).
1.2 NE U R O N A L E N C O D I N G
Specific modalities of sensory stimuli such as pressure, sound level or visual
contrast have been thought to be encoded by the spiking rate of neurons. In the visual
system, the relay cells in the LGN transmit visual information encoded by retinal
ganglion cell responses to the primary visual cortex. The notion of feature encoding by
firing rate has its roots in the seminal studies of Hubel and Wiesel (1961) using
simplified stimuli such as light spots and oriented bars.
A simple demonstration of how the spiking responses encode stimulus
characteristics can be seen in the experiments of Hubel and Wiesel (1961) in the LGN.
As shown in Figure 2, the size of the stimulus strongly modulates the neuronal
responses. When the stimulus is relatively large but still confined to the the center of the
Figure 1. Schematic representation of a cat LGN in a coronal section showing its laminar organization in three main layers, A, A1 and C, and the lines of projection representing single points in the visual field. OT, optic tract (from Sherman and Guillery, 2002).
Fast osci l lat ions in the ret ina by Fabio Freitag 11
RF, the responses are vigorous. However, when the stimulus reach the surround
subregions of the RF, responses decreases. Thus, the concentric organization of the RFs
in the retinogeniculate pathway provides a way of encoding a simple feature of the
visual stimulus, its size.
Average firing rates, however, are not the only possibility for representing a
stimulus. It has been argued that the fine temporal structure of the responses may be
relevant for encoding. Accordingly, experimental and theoretical evidence has shown
that the relative timing structure of single cell responses (Reich et al., 1997), relative
spike latencies in populations (Gollisch and Meister, 2008), and phase-locking between
spikes and ongoing oscillatory activity (Koepsell et al., 2010) carry information about
the stimulus.
Figure 2. Stimulus size modulates spiking responses of an ON-center cell in the LGN. When a circular stimulus covered entirely the center of the RF responses reach their maximum. Notice that firing rate decreases for a very large stimulus covering the surround region of the RF. Responses are poor when there is no light in RF center (modified from Hubel and Wiesel, 1961).
Fast osci l lat ions in the ret ina by Fabio Freitag 12
1.3 TH E B I N D I N G B Y S Y N C H R O N I Z AT I O N H Y P O T H E S I S
As discussed before, information from the visual scene is decomposed and
processed in parallel pathways which ultimately converges onto higher visual areas,
where neurons respond selectively for combinations of color, motion, orientation,
texture, shape and depth. Feedforward connections from earlier processing stages to
higher areas, such as V2, V4 and MT, coexist with reciprocal feedback (Rosa and
Tweedale, 2005; Sillito et al., 2006) and lateral connections (from equivalent level of
processing) (Wunderle et al., 2013). Thus, the visual system seems to work in a
hierarchical fashion, first analyzing the various elemental features which are combined
later in higher areas. How bits of visual information are bound into coherent wholes?
Which are the mechanisms responsible for perceptual binding? This is still a largely
unresolved question and still a big challenge for neuroscience.
An important step for understanding perceptual binding mechanisms was made
by the synchronization hypothesis proposed by von der Malsburg (1981). Essentially,
Christoph von der Malsburg‘s idea was that the elementary features of perceptual
objects are linked together by the synchronous activity of distributed neuronal
assemblies. The synchronization of the spiking activity at very fine time scale (order of
milliseconds) would provide a mechanism for building flexible relationships in the
brain. This would explain how we bind information related to one object and distinguish
from the background (see the dalmatian dog in Figure 3).
Figure 3. The binding by synchronization hypothesis. Local features in a visual scene (spots of the dalmatian dog) are linked by the synchronous activity among neuronal assemblies. In accord to this hypothesis, visual objects (the whole dog) are segregated from the background because of temporal relationships created in the brain. Local field potential (LFP) and spiking activity (MUA) traces show synchronous oscillatory responses in the visual cortex in response to a single coherent bar (data from Gray and Singer, 1989).
Fast osci l lat ions in the ret ina by Fabio Freitag 13
Experimental evidence for the binding by synchronization hypothesis was first
obtained in the visual cortex of the cat (Eckhorn et al., 1988; Gray et al., 1989). In
response to simplified stimuli, such as moving bars, single cells and groups of cells
(multi-unity activity, MUA) in areas 17 and 18 exhibit strong oscillatory activity at the
gamma range, from 35 to 85 Hz (Figure 3). In these seminal studies, periodic activity in
spiking activity was found to be highly correlated with the simultaneously recorded
local field potential (LFP). Action potentials generated synchronically within gamma
cycles would have a higher probability to generate new action potentials in target cells,
allowing for a neuronal coding based on timing relationships (see review in Uhlhaas et
al., 2011). These findings suggest that columns at different regions of the cortex may
synchronize their activity, leading to synchronous ensembles representing distinct
surfaces or objects in a visual scene. Interestingly, gamma oscillatory responses were
not found in the LGN (Gray and Singer, 1989) indicating that the gamma oscillations
resulted from an intracortical mechanism.
An important finding in these early studies was that synchronization of neuronal
responses appears only for single contour objects (such as a single coherent bar, Gray et
al., 1989) but not by different contours moving in different directions (Engel et al.,
1991), revealing that synchronization is highly context-dependent (Singer, 1999). The
idea that synchronization can mediate a binding mechanism was supported later by a
series of experimental studies in cats and monkeys (Engel et al., 1991; Kreiter and
Singer, 1996). Evidence was obtained not only for simple moving bars but also for bi-
stable stimuli such as moving plaids (Castelo-Branco et al., 2000). Cells in area 18 and
another cortical area called postero-medial bank of the lateral suprasylvian sulcus
(PMLS) of the cat synchronize their responses when induced by contours of the same
surface (pattern motion of the plaids) but not by contours belonging to different surfaces
(component motion of the plaids). These results are very interesting because in principle
they could provide an explanation for perceptual dynamics.
A number of studies, mainly in behaving monkeys, however, have shown
contrary evidence for the binding by synchronization hypothesis (Thiele and Stoner,
2003; Roelfsema et al., 2004; Palanca and DeAngelis, 2005; Lima et al., 2009). These
studies are very important in trying to establish a direct correlation between neuronal
synchronization and perception, what was not possible in the early studies in the
Fast osci l lat ions in the ret ina by Fabio Freitag 14
anesthetized cat. In addition, a body of evidence has been gathered recently showing
that gamma processes are associated with attention (Engel et al., 2001; Fries et al.,
2001, see review in Fries, 2009) and temporal expectancy (Lima et al., 2011). These
latter studies suggest that gamma synchronization may be important for controlling the
flow of information across the cortical networks.
Overall, a consensus on the role of gamma oscillations in cognitive function is
still lacking. A possible reason is that gamma oscillations may be related to more than
one elementary function in the highly diverse neural subsystems that composes the
brain (Merker, 2013).
1.4 SY N C H R O N O U S O S C I L L AT I O N S I N T H E E A R LY V I S U A L S Y S T E M
Oscillations are not restricted to the cortical networks. In fact, oscillatory
responses were observed in the optic nerve and eye as early as the first recordings of
Figure 4. Fast retinal oscillations have been described in a variety of animals, such as the frog, rabbit (not shown), cat and monkey, with intracellular (frog) and extra-cellular recording techniques. Oscillations are generally seen in MUA responses to large visual stimuli and also spontaneously, in the dark.
A N E S T H E T I Z E D M O N K E Y
A N E S T H E T I Z E D C AT
B E H A V I N G F R O G
Doty and Kimura, 1963
Laufer and Verzeano, 1967
Neuenschwander and Singer, 1996
Doty and Kimura, 1963
Ishikane et al., 2005
Fast osci l lat ions in the ret ina by Fabio Freitag 15
neuronal activity became available (Gotch, 1903; Fröhlich, 1914; Granit and Therman,
1935). Activity in sensory systems, in particular, are commonly associated with fast
oscillations, such as in the olfactory bulb (Laurent, 2002), somatosensory (Ahissar et
al., 2000) and auditory cortices (Roberts and Rutherford, 2008), and in the insect optic
lobe (Kirschfeld, 1992). In the retina, prominent oscillations have been described in
various species, including the frog (Ishikane et al., 1999; Ishikane et al., 2005), rabbit
(Ariel et al., 1983), cat (Doty and Kimura, 1963; Laufer and Verzeano, 1967; Arnett,
1975; Neuenschwander and Singer, 1996; Neuenschwander et al., 1999) and monkey
(Doty and Kimura, 1963) (Figure 4). Strong oscillations in retinal ganglion cell activity
can be observed for responses to large spots of light, covering the center and the
surround subregions of the RFs (Neuenschwander et al., 1996), as well as Ganzfeld
illumination (Laufer and Verzeano, 1967), and also spontaneously, in the dark (Bishop
et al., 1964; Neuenschwander et al., 1999). Correlation analysis of multi-electrode
recordings suggested that synchronous oscillations in the retina arises from large-scale
interactions between the ganglion cells (Laufer and Verzeano, 1967; Neuenschwander et
al., 1999, see example in Figure 5). In the cat, synchronization of oscillatory responses
has been observed for distances larger than 20 degrees of visual angle across the retina
(Neuenschwander and Singer, 1996). Moreover, fast oscillations were found in
Figure 5. Synchronous oscillations in the retina arises from population interactions. MUA traces show strong oscillatory responses to a large spot of light. Notice that correlated individual spikes from different cells exhibit missing cycles. Color arrows indicated spikes from different cells. Qualitative analysis (data from Neuenschwander and collaborators, Max-Planck Institute for Brain Research, Frankfurt).
10 ms
30 ms
Fast osci l lat ions in the ret ina by Fabio Freitag 16
responses to all functional types (ON and OFF-cells, X- and Y-cells) in the retina and
LGN (Ariel et al., 1983; Neuenschwander et al., 1999; Ito et al., 2010).
An important aspect in the organization of the retinogeniculate system is the
reliability of information transfer. Retinal afferents to the LGN consist of thick axons
involving richly branched terminal arbors with boutons densely distributed in terminal
clusters (Sherman and Guillery, 2006). This exquisite organization confers great
robustness for the retinogeniculate transmission. Analysis of retinal excitatory post-
synaptic potentials (EPSPs) associated with LGN spike waveforms (S-Potentials) shows
that most LGN neurons have one retinal ganglion cell input that accounts for nearly all
LGN spikes sent to visual cortex (> 95% of LGN spikes are associated with an EPSP
from a single retinal ganglion cell, Sincich et al., 2007). Thus, it is not surprising to see
that the retinal fast oscillations are faithfully transmitted through the retinogeniculate
pathway up to the LGN. Accordingly, it has been demonstrated by means of
simultaneous recordings from the retina and LGN that retinal oscillatory inputs survive
transmission at the retinogeniculate synapses (Arnett, 1975; Neuenschwander and
Singer, 1996). Moreover, simultaneous recordings from the LGNs in the two
hemispheres have shown that only LGN cells receiving inputs from the same eye are
Figure 6. Feedforward synchronization in the retinogeniculate system. Synchronization of oscillatory responses is seen only among LGN cells receiving inputs from the same eye (right lamina A and left lamina A1, RA - LA1), whereby it did not matter whether the cells were located in the same LGN or in the LGNs of the two hemispheres. Direct recordings form the retina and LGN have demonstrated that the synchronizing mechanism resides in the retina (from Neuenschwander and Singer, 1996).
LEFT EYE
LEFT LGNRIGHT LGN
RARA1 LA1
-80 0 80Time lag (ms)
100
Fast osci l lat ions in the ret ina by Fabio Freitag 17
capable of synchronizing their responses (Figure 6), excluding thalamic mechanisms
since they cannot account for synchronization across the two hemispheres
(Neuenschwander and Singer, 1996). In addition, recordings from the retina, LGN and
visual cortex in the cat (Castelo-Branco et al., 1998) have provided direct evidence that
oscillatory responses from the retina are transmitted to the cortex, characterizing a
feedforward synchronization mechanism in the early visual system (Neuenschwander et
al., 2002).
It is important to notice that although retinal oscillations are transmitted to the
cortex, they do not necessarily contribute to generate cortical gamma. In the study of
Castelo-Branco et al. (1998) in the cat, data from simultaneous recordings from the
retina, LGN and cortex (areas A17 and A18) indicate that gamma in the cortex arises
independently from oscillatory retinogeniculate inputs. Moreover, in the cortex, gamma
responses are very sensitive to the orientation selectivity of the cells (Gray and Singer,
1989), a feature that is irrelevant for responses in the retina. Another important finding
in the study of Castelo-Branco et al. (1998) is the noticeable differences in oscillation
frequencies. Gamma oscillations in the cortex have frequencies in the range of 30 to 60
Hz, while fast oscillations in the retina have frequencies at a much higher band, from 60
to 120 Hz. It is possible, however, that synchronized responses in the retina bias
synchronization of responses at the cortical level. Alternatively, synchronous inputs to
postsynaptic cells may improve information transmission (Dan et al., 1998). For all
these conjectures, at present it is still unclear whether fast oscillatory retinogeniculate
inputs contribute or not to cortical processing.
1.5 RE T I N A L O S C I L L AT I O N S A S A N E N C O D I N G M E C H A N I S M
If retinal oscillations indeed play any role in vision they most likely convey
stimulus information upstream to the cortex. In the anesthetized cat, Neuenschwander
and Singer (1996) showed evidence that synchronization of oscillatory responses in the
retina (and in the LGN) depends on stimulus continuity. Rectangular stimuli flashed
over the RFs elicited strong oscillatory responses, which were correlated only when the
stimulus was contiguous, linking the two RFs (Figure 7). Based on these findings, the
authors proposed that retinal oscillations serve as a mechanism for conveying
information about stimulus connectedness or continuity. This interpretation has a
Fast osci l lat ions in the ret ina by Fabio Freitag 18
parallel with the hypothesis of feature binding by gamma synchronization. Cortical
feature space is far more complex, such as the representation of stimulus orientation,
color, form and texture. In the retina, on the other hand, representational space is
restricted to simpler features, such as position, contrast and size. These ideas gained
support with a recent model of the inner retina showing that correlated oscillatory
activity in the retina convey size information (Stephens et al., 2006). It is possible that
large features in a visual scene, such as continuous surfaces, are encoded by the global
synchronous population activity in the retina. To determine if encoding of stimulus size
depended on connectedness independent of object shape, Stephens et al., (2006) ran a
simulation of responses to a random binary checkerboard stimulus, in which every pixel
changed incrementally its color from black to white (see example in Figure 11). The
sudden appearance of large connected blobs in the stimulus (at the percolation
threshold) led to a sharp phase transition in the oscillatory behavior. Interestingly,
synchronization of oscillatory responses of retinal ganglion cells in the frog is correlated
with escape behavior (Ishikane et al., 2005). These findings are relevant because they
point to a direct link between retinal oscillations and behavior.
Encoding by temporal correlations and by firing rates are not exclusive. As
proposed by Koepsell et al. (2009), a temporal code may co-exist with a rate code in the
Figure 7. Synchronous oscillations in the retina encode stimulus continuity (from Neuenschwander and Singer, 1996). Response to rectangular light stimuli that were either separated by a gap or were contiguous and covered the area between the RFs. When the stimuli were separated, they elicited responses that oscillated with similar frequency (ca. 110 Hz) at both sites but showed no synchrony (left). When the stimuli were spatially contiguous, the oscillatory responses in the LGN became highly synchronous (right).
Fast osci l lat ions in the ret ina by Fabio Freitag 19
retina. It is even possible that information about local contrast features and global size
are carried independently by multiplexed channels.
In the present study, we investigate the role of fast oscillations in the retina on
stimulus encoding based on recordings from the LGN in the anesthetized cat. We used a
variety of simplified stimuli to test how oscillatory behavior is related to stimulus
continuity and connectedness. In particular we tested experimentally the stimulus
proposed in the theoretical study of Stephens et al. (2006). We also used dynamical
stimuli, such as natural scene movies, to verify whether retinal oscillatory responses are
present for more naturalistic conditions.
1.6 AN E S T H E S I A E F F E C T S
Halogenated anesthetics, such as halothane and isoflurane, have been commonly
used in many studies of rhythmic activity in the brain. In the cat, cortical gamma
oscillations seem to survive the anesthesia, although efforts are generally made to have
it at low levels (lightly anesthetized animal, e.g., Gray et al., 1989). Nevertheless, it has
been demonstrated with stimulation of the mesencephalic reticular formation that high
cortical activation levels (Munk et al., 1996; Herculano-Houzel et al., 1999) facilitate
gamma activity in the cortex of anesthetized cats. Moreover, a series of recent studies in
the behaving monkey have shown that gamma responses are modulated with attention
(Womelsdorf et al., 2007; Fries 2009) and temporal expectancy (Lima et al., 2011).
Based on these reports one may conclude that low cortical activation, as during the
anesthesia, generally damps gamma responses in the cortex.
Similarly, retinal oscillations have been studied mostly during halothane or
isoflurane anesthesia (Laufer and Verzeano, 1967; Neuenschwander and Singer, 1996;
Ito et al., 2010). A few studies in the 60‘s reported retinal oscillatory responses in non-
anesthetized, paralyzed cats submitted to transpontine transection (Doty and Kimura,
1963). To clarify whether halothane has an effect on the generation of retinal
oscillations we varied systematically the concentration levels of this anesthetic during
the experiments.
Fast osci l lat ions in the ret ina by Fabio Freitag 20
2. OB J E C T I V E S
Our primary goal with this study was to test if retinal oscillations (recorded here
in the LGN) serve as an encoding mechanism for stimulus size and continuity as
proposed by a number of theoretical studies. In order to investigate stimulus
dependencies we applied a series of stationary stimuli of various sizes and luminance
values. We compared the oscillatory responses to these simplified stimuli with
responses to stimuli of increasing complexity, such as binary and gray-level natural
movies.
In addition, we have examined how oscillatory responses recorded in the LGN
are dependent on the anesthesia. We studied in particular the effects of halothane and
isoflurane, halogenate anesthetics that have been routinely used in many studies of
neuronal oscillations in cats and monkeys.
Specif ic goals
• Evaluate the dependencies of retinal synchronous oscillations on stimulus size
and contrast;
• Verify whether breaking of stimulus continuity disrupt synchronous oscillatory responses;
• Test if the sudden emergence of large connected clusters (percolation threshold in random binary checkerboards) leads to transitions in the oscillatory patterning of the responses;
• Verify whether retinal oscillations are also present in responses to spatial and temporal complex stimuli, such as gratings and natural scene movies;
• Test the dependencies of the oscillatory responses on halothane (and isoflurane) anesthesia levels.
Fast osci l lat ions in the ret ina by Fabio Freitag 21
3. ME T H O D S
Adult cats from the Brain Institute’s colony were used in this study (N= 5). All
experimental procedures were previously approved by the ethical committee for animal
experimentation of the Universidade Federal do Rio Grande do Norte (CEUA-UFRN
protocol nº 019/2012).
Recordings of LGN responses in anesthetized cats were obtained by means of up
to 5 electrodes placed at the representation of the central visual field. Our analysis
consisted in obtaining indicators for the incidence and strength of oscillatory responses
in the retina, which are known to be transmitted faithfully to the LGN (Neuenschwander
et al., 2002). Auto and cross-correlation functions were computed for spiking responses
in order to estimate the modulation amplitude and frequency of the oscillations. The
time course of the oscillatory responses was evaluated with a sliding window analysis.
In addition, the rate levels were estimated by means of average response histograms
(PSTHs). Analysis was based on the NEUROSYNC package developed by S.
Neuenschwander, at the Department of Neurophysiology, Max-Planck Institute for
Brain Research, Frankfurt.
To test the role of retinal oscillations in stimulus encoding we used a variety of
visual stimuli, including spots of light of various sizes and contrasts, checkerboards,
moving gratings and natural scene movies.
In addition in this study we examine the effects of halogenated anesthetics such
as halothane on retinal oscillations.
3.1 EX P E R I M E N TA L S E S S I O N S
Experiments were done in the anesthetized cat. Different anesthetics (e.g.,
ketamine, halothane, isoflurane, fentanyl) were used to establish whether retinal
oscillations are dependent on the agent and level of anesthesia.
3.1 .1 SU R G I C A L P R O C E D U R E S , A N E S T H E S I A A N D L I F E S U P P O RT
An experiment started with an intramuscular injection of ketamine (10 mg/kg,
i.m.) and xylazine (2 mg/kg, i.m.). The cats were then submitted to a tracheotomy to
Fast osci l lat ions in the ret ina by Fabio Freitag 22
allow the maintenance of the inhalation anesthesia by artificial ventilation with a gas
mixture of N2O (70%) and O2 (30%), supplemented with halothane (0.8 to 1.2 %,
Tanohalo - Halotano, Cristália; in some experiments isoflurane was also used at similar
concentrations). To ensure the immobilization of the eyes, a muscle relaxant
(pancuronium bromide, 0.25 mg/kg/h) was administrated i.v. continuously during the
experiment. Vital signs such as the ECG, ventilation pressure, rectal temperature
(maintained at 37°C) and expired CO2 (kept in the range of 2.6 to 3.5 %) were
continuously monitored during the experiments by means of a medical life monitoring
system (GE Dash 3000 Patient Monitor). Fluid loss was compensated by continuous i.v.
infusion of saline solution and glucose (5%).
After the animal was placed in the stereotaxic apparatus, a craniotomy was made
in the skull above the LGN (AP 6.5 and ML 9.5). Subsequently, a recording chamber
made of titanium with 14 mm of internal diameter was mounted over the exposed
region and secured into the bone with titanium screws and dental cement. A metal rod
was also secured on the skull for fixation of the head. At the end of all surgical
procedures, the stereotaxic ear and eye bars were removed and the halothane
concentration adjusted between 0.6 and 0.8%. Recordings were obtained during the
following 3–5 days.
During visual stimulation, the pupils were dilated and the nictitating membranes
retracted by means of atropine applied topically (1%). Contact lenses with an artificial
pupil of 3 mm diameter were used to protect the cornea. The representation of the area
centralis was plotted on a white board with help of a fundus camera (Zeiss, Germany)
and served as a reference for estimating the position of the receptive fields.
3.1 .2 RE C O R D I N G D E V I C E A N D E L E C T R O D E S
For the recordings we used a custom made system fitted with up to 5
independently movable electrodes. This system was designed by S. Neuenschwander at
the Max Planck Institute for Brain Research, Frankfurt.
The recording device consisted of 5 Narishige hydraulic microdrives mounted in
a movable platform made in thermoplastic (PEEK) (Figure 8). An acrylic plate secured
Fast osci l lat ions in the ret ina by Fabio Freitag 23
a 6 X 5 grid for positioning of the guiding-tubes. The position of the guiding-tubes was
decided previously based on stereotaxic atlas of the cat.
For electrode guiding two needles were mouted in each other (Braun 100
Sterican Ø 0.60 x 60 mm, 23G x 2 ⅜” and Ehrhardt Supra Einmal-Kanülen Ø 0.30 x 23
mm) to obtain a total length of 82 mm. The thicker extremity was used for fixation in
grid (the hole size was sufficient to allow the passage and hold the guiding-tubes), while
the thinner one was placed into the brain.
A glass capilar mounted at the end of the guiding-tubes served as a piston for
moving the electrodes. A piece of needle was used to connect the glass capilar to the
micromanipulators. The recording device was fixated on recording chamber with an
adapter screwed in the recording chamber. X-Y positioning of the guiding-tubes was
adjusted previously to the descent of the electrodes by means of a Narishige X-Y table
(parts of the Narishige MO-95 recording system).
In this study we used tungsten-platinum, quartz-insulated, fiber electrodes (80
µm diameter, Thomas Recording, Germany), see Figure 9. These electrodes were
Figure 8. Schematic representation of our electrode recording system based on Narishige oil-hydraulic microdrives and X-Y table (parts of the MO-95 microdrive, Narishige, Japan). Designed by S. Neuenschwander at the Max-Planck Institute for Brain Research, Frankfurt, 2005. Simulation by Heitor de Oliveira, Brain Institute - UFRN, Natal.
Macro-screw (for macro-movement of the electrodes)
Narishige MO-95 micromanipulators (fro micro-movement of the electrodes)
Electrode macro depth scale (macro-movement of the electrodes)
Guiding-tubes gridGuiding-tubes
Device X-Y table
Narishige micromanipulators sliding suport
Guiding-tubes depth scale Guiding-tubes sliding suport
Fast osci l lat ions in the ret ina by Fabio Freitag 24
chosen because they have a very good signal to noise ratio, they are strong enough to
penetrate the intact dura-mater, and have high biocompatibility. The tips were prepared
in a grinding machine (Thomas Recordings, Gießen, Germany) to expose the metal core
with a conical shape (expose tip of 20 µm). Electrodes were connected by welding thin
flexible wires to them.
3.1 .3 S I G N A L C O N D I T I O N I N G A N D D ATA A C Q U I S I T I O N
After the fixation of the recording device and positioning of the guiding-tubes
we generally made a series of exploratory penetrations, usually with a single electrode,
in order to find an appropriate position of the electrodes aiming at the central
representation of the visual field.
Spiking activity from neuronal groups (MUA) were obtained after amplification
and band-pass filtering the signals (0.7 – 6.0 KHz) with a 32-channel Plexon modified
preamplifier fitted with an HST16025 headset (Plexon Inc, Dallas, TX, USA). Data
acquisition was made using the program SPASS, an acquisition system based on
National Instruments boards (M-series acquisition boards, National Instruments, Austin,
TX, USA) written in LabVIEW by S. Neuenschwander at the Max Planck Institute for
Brain Research, Frankfurt. Signals were sampled at 32 kS/s with an additional 10 X
onboard amplification. Spikes were detected with an amplitude discrimination with
threshold set to twice the noise level.
Figure 9. Quartz-insulated tungsten-platinum electrodes. Notice that electrode‘s impedance depends largely on the size of the exposed tip. Electrodes of 0.5 Ω impedance as used in this study are better for MUA recordings. Electrodes were manufactured in our laboratory from quartz electrode fibers using a a special grinding machine (Thomas Recording, Gießen, Germany).
80 µm
METAL CORETIP
1.0 MΩ
0.5 MΩ
QUARTZ
Fast osci l lat ions in the ret ina by Fabio Freitag 25
3.2 . V I S U A L S T I M U L I
Visual stimuli were presented on a 21” CRT monitor (Hitachi, CM815ET),
placed at 57 cm of distance from the cat, with a refresh rate of 100 Hz and 1024 x 768
pixels resolution. In these conditions 1° of visual field corresponded to approximately to
25 pixels. The screen was gamma-corrected in order to achieve a linear perception of
luminance (illumination range from 25 to 200 Lux). To control the stimulus presentation
we used the software ActiveStim (www.activestim.com). Protocols consisted of a series
of 60 to 300 repetitions. Generally, stimuli were centered at the mapped receptive field.
All different types of stimuli from different protocols were presented in a random order.
At the beginning of the recording sessions, the receptive field of the cells were
mapped with an automatic algorithm based on a moving bar (1000 X 5 pixels) crossing
the whole field of the monitor in 16 different directions (Pipa et al., 2012). RF maps
were obtained by compiling an average matrix, in which the responses to the moving
bar were added in 10 ms bins (corresponding to 0.2° in visual angle) for all directions
(see example in Figure 14). Depending on the eccentricity of the RFs the electrodes
were moved to a new position aiming at a region corresponding to the central
representation of the visual field.
Figure 10. Stimulus size and circle annular mask protocols. We used annular masks to test whether breaking stimulus continuity disrupts oscillatory responses to a circle. Stimuli were centered at the RF fields. Stimulus sizes are indicated in degrees of visual angle (same conventions in all other figures).
C I R C L E S I Z E
C I R C L E A N N U L A R M A S K
3° 7° 9° 9° 10° 10° 12°12°5° 7° 14°
10°2° 3° 4° 6° 7° 8° 11° 12° 13°
14°
Fast osci l lat ions in the ret ina by Fabio Freitag 26
Circle size
In order to analyze the effects of the stimulus size and contrast on firing rates
and oscillatory responses, we presented circles of 10 different sizes (2°, 3°, 4°, 6°, 7°,
8°, 10°, 11°, 12° and 13° of visual angle) with 3 different contrast levels (relative levels
of 1.0, 0.5 and 0.3, ranging from 175 to 50 lux; background illumination of 1.0 lux).
Sessions comprised 10 stimulus repetitions for a total of 300 trials.
Circle annular mask
We used annular masks to test whether breaking stimulus continuity disrupts
oscillatory responses to a circle. These stimuli had a shape of donut and were placed
centered over individual RFs, in a way that the responses to the central circle were
independent of the peripheral annulus. Annular masks and solid circles were presented
randomly with different sizes (sequence of 3° circle, 5° circle, 7° circle, 7° masked
circle, 9° circle, 9° masked circle, 10° circle, 10° masked circle, 12° circle, 12 ° masked
circle, 14° circle and 14° masked circle; Figure 10).
Percolation stimulus
Our percolation protocol consisted in presenting sequences of checkerboard
binary images, in which every pixel changed its luminance randomly and once, from
black to white (gray background, see Figure 11). This stimulus was proposed by
Stephens and colleagues in a theoretical study (Stephens et al., 2006) to test if
oscillation dynamics in the retina show a transition point in response to a change in the
spatial properties of the images, when all points get connected (percolation phase
Figure 11. Percolation stimulus. This stimulus protocol was designed to test if oscillation dynamics in the retina show a transition point in response to a change in the spatial properties of the images, when all points get connected (percolation threshold). RF position is indicated by the color circle in the center of the stimulus (same conventions in all other figures).
GRATINGS
Fast osci l lat ions in the ret ina by Fabio Freitag 27
transition). The duration of this protocol was controlled and ranged from 1 to 6 seconds
and each session had either 50 or 100 trials. Thus, the speed of the appearing white
pixels also changed. As a control, we used a solid circle (of same size) with luminance
increasing linearly (ramp). This control stimulus was presented randomly with the
percolation stimulus.
Natural movies and gratings
In general, retinal oscillations have been studied with simplified stimuli, such as
spots of lights, bars, gratings, and Ganzfeld stimuli (Neuenschwander et al., 1999;
Laufer and Verzeano, 1967). In many cases the stimuli were static and uniform, very
different from the natural vision. To test how spatial and temporal dynamics in the
stimulus change oscillation behavior we used 4 types of dynamic stimuli: gray-scale
natural scene movies (200 luminance values), binary natural scene movies (2 luminance
values), randomly flickering checkerboards (temporal frequency from 25 to 50 Hz) and
moving gratings (temporal frequency 0,33 Hz, 1,25 Hz and 3 Hz) (see examples in
Figure 18). Two different natural scene movies were presented (referred here as
LabPan and Medikament), consisting of pan images of the laboratory (the same stimuli
Figure 12. Dynamical visual stimuli. Gray-level natural scene movies had 200 luminance values (LabPan movie is depicted here). Binary natural scene movies were displayed with 2 luminance values (black and white). The movies were made with a single global motion component made with a long camera panning (same stimuli used in Haslinger et al., 2012). Random stimuli consisted of flickering checkerboards. Grating stimuli moved at temporal frequencies of 0,33 Hz, 1,25 Hz and 3 Hz.
RANDOM
GRATINGS
GRAY-LEVEL BINARY
Fast osci l lat ions in the ret ina by Fabio Freitag 28
used in the study of Haslinger et al., 2012). Trials had duration of 5 sec and each
stimulus was repeated 60 times in a pseudorandom order.
3.3 . HA L O T H A N E E F F E C T S
To test the effects of halothane anesthesia we varied systematically its
concentration levels during the recording sessions (Dräger Vaporizer, Germany).
Accurate estimation of the partial expiratory concentration of halothane (and isoflurane)
in the ventilation system was obtained from readings of the life monitoring apparatus
(GE Dash 3000 Patient Monitor). During the manipulations of halothane (and
isoflurane) levels we supplemented the anesthesia with ketamine or fentanyl (even in
case of total absence of halothane or isoflurane). These supplementary anesthetics were
applied well before the recording sessions.
For testing the effects of halothane we used various static and dynamical stimuli
(circle, linear ramp, natural scene movies and gratings) with characteristics known to
induce strong oscillatory responses in the retina (large, high contrast stimuli). For the
circle stimuli, halothane concentration was changed from 0.74% to 0.20% (at 20 trials
recording sessions for each condition), or smoothly from 1.0% to 0.0% (300 trials
recording session over about 1 hour). For the linear ramp stimuli, we decreased the
halothane levels from 0.95% to 0.20% to obtain recording sessions of 50 trials (2.0 to
3.0 sec duration). For the natural scene movies, we decreased the halothane levels from
1.0% to 0.19% in 3 steps. Finally, for the gratings stimuli, we decreased the halothane
levels from 0.94% to 0.0%. In all cases as a control the concentration of halothane was
restored to previous levels. In a few experiments we substituted halothane for
isoflurane, which is of the same family of halogenated anesthetics.
3.4 . DATA A N A LY S I S
Our analysis was mainly focused on the spiking activity of groups of neurons
(multi-unity activity, MUA). Only occasionally we have used spike sorting methods to
study single unity activity (SUA). Analysis was made using the NEUROSYNC
packaged developed in LabVIEW (Austin, Texas, USA) and in Matlab R10
(MathWorks, Natick, Massachusetts, USA). For spike sorting we used the program
Fast osci l lat ions in the ret ina by Fabio Freitag 29
Wave_clus developed by Rodrigo Quiroga (Quiroga et al., 2004). This method uses a
set of wavelet coefficients from each spike selected by a Kolmogorov-Smirnov test of
normality as an input to the clustering algorithm. An automatic clustering algorithm
(super-paramagnetic clustering) identifies different clusters of spikes based on the
selected set of wavelet coefficients.
In order to access the rate and timing of neuronal spike activities induced by
visual stimuli, response histograms were computed with a resolution of 10 ms. We
decided to use small bin sizes because retinal oscillations are known to be very fast (up
to 100 Hz). A 500 to 1000 ms window was used for estimating average firing rates over
different response epochs.
Our analysis of neuronal oscillations was made in the time domain. Correlation
functions for spiking responses were computed for 500 to 1000 ms analysis windows.
For a quantitative analysis a Gabor function was fitted to the correlograms and its
parameters were used to infer the strength and frequency of the synchronous
oscillations. The Levenberg-Marquardt algorithm was used (described in Press et al.,
1986) for the fitting procedure. To estimate correlation strength, a modulation amplitude
index was computed as the ratio between the amplitude of the central peak of the fitted
function and its offset. In cases when an oscillatory patterning of the responses was
detected, we used two metrics to evaluate oscillations strength: the oscillation amplitude
(correlograms were normalized by the geometric mean of firing rates) and the
modulation amplitude of the first satellite peak (MAS) (König, 1994).
A sliding window analysis was applied to follow the oscillatory behavior of the
responses over time. This analysis consists of sampling the response using a 100-ms
analysis window in 50-ms steps. Correlograms obtained for each overlapping window
were then plotted as a two-dimensional array, where the y-axis denotes time lag of the
correlation and the x-axis denotes the time course of the responses. The amplitude of the
correlograms was normalized by the geometric mean of firing rates and displayed with a
color code.
Comparisons between stimulus conditions group data was subject to the analysis
of variance (ANOVA, after testing for normal distribution) and post-hoc Bonferroni for
testing differences in pairs of mean values. Confidence interval was chosen with 95% (p
< 0.05).
Fast osci l lat ions in the ret ina by Fabio Freitag 30
4. RE S U LT S
4.1 . DE P E N D E N C I E S O N S T I M U L U S S I Z E
To test how stimulus size modulates the strengh of retinal oscillations (seen here
in LGN responses) we used a simple circular light stimulus, which was flashed over the
RFs of the recorded neurons. The sliding analysis presented in Figure 13 gives an
example of how strong oscillations in the retinogeniculate system can be in response to
a large stimulus (in this case, 10°). We used 10 different sizes of the simulus at 3
luminance levels (175, 80 and 50 Lux, corresponding to contrast ratios of 1.0, 0.5 and
0.3) (N= 14 recording sessions, comprising 3 different experiments, 10 repetitions for
10 different sizes). As a metric to characterize the oscillation strength in MUA
responses, we have taken the amplitude of the oscillation from the fitted function in the
correlograms (oscillation amplitude, see Methods). Comparisons across conditions have
shown significant differences for stimulus size for all contrast levels (p < 0.05 for the
stimulus sizes of 4°, 7°, 8°, 10°, 11° and 12°). Mean values and standard deviations for
oscillation amplitude and firing rates are given in Table 1.
To analyze how oscillation strength varied as a function of stimulus size we
compiled plots separately for the 3 luminance values studied (Figure 13). Oscillation
strength increased as a function of stimulus size non linearly, without reaching a
plateau. The level of responses, on the other hand, reached a maximum already for a
stimulus of 4°. Thus, in our data, oscillation strength was clearly independent of the
spike rate.
Synchronization in the LGN was always accompanied by oscillations. Figure 14
shows an example of a cross-correlation computed for MUA pairs with RFs separated
by about 3°. Recordings were obtained from different electrodes. Observe that the
modulation of the correlogram is very strong (oscillation frequency, 87 Hz; MAS, 0.76).
The shift-predictor control is flat indicating that oscillatory responses do not arise from
Fast osci l lat ions in the ret ina by Fabio Freitag 31
Figure 13. Stimulus size and luminance modulate fast oscillations in the retinogeniculate system. MUA recordings were obtained from the LGN of the anesthetized cat. (Left panels) Autocorrelation sliding window analysis show strong oscillatory responses to a high contrast circle of 10° diameter flashed over the RF (oscillation frequency at 78 Hz), while a 8° stimulus evoked weak oscillations for the same contrast. Sliding window size, 200 ms; step, 100 ms. (Right plots) Population data, traces show response levels (firing rate, upper plot) and oscillation amplitude (lower plot) as function of stimulus size. Notice that rates and oscillation strength have different trends. Plots show data obtained for 3 different stimulus luminance levels (175, 80, 50 Lux). Analysis window of 1000 ms, as indicated by the dashed line at the bottom of the sliding window plots. In this study correlograms were normalized by firing rates (normalized coincidences). Arrows indicate the points corresponding to the data used for the sliding window analysis (corresponding respectively to stimulus sizes of 8° and of 10°).
0.25
175 Lux
50 Lux80 Lux
2° 3° 4° 6° 7° 8° 10° 11° 12° 13°
Rat
e (s
p/s)
Osc
. A
mpl
itud
e
C i rc le s ize
0.00
200
0
Tim
e (m
s)
80
0
-80
Tim
e (m
s)
80
0.3
0.0
0
-80
Co
inci
den
ces
8°
10°
Time5 0 0 m s
cgl04i0206 5-5
cgl04i0207 5-5
Figure 14. Synchronous oscillatory responses to a circular light stimulus for a pair of MUA ON-cells with corresponding RFs separated by approximately 3°. Panels show crosscorrelograms computed with a bin of 1 ms. Oscillation frequency, 87 Hz; Modulation amplitude (MAS), 0.76. Shuffling the correlation by 1-trial (shift-predictor) gives a flat function, indicating that correlation arises from interactions in the retinal networks. Fitted functions are displayed as black thin lines over the correlograms. Analysis window, 1000 ms. Stimulus was flashed over the RFs (size, 14°). (Upper inset) RF mappings obtained by means of a moving bar showed at 16 different directions (Pipa et al., 2012). Data from two independent electrode channels (RF maps were merged for visualization). Stimulus borders are indicated by the circle line on the RF mapping plots.
2.0
0.0Coi
ncid
ence
s
-80 0 80Time (ms)
CROSSCORRELATION
2°cgl03d0513 5-4
87 Hz0.76
14°
CONTROL
Fast osci l lat ions in the ret ina by Fabio Freitag 32
stimulus coordination.
4.2 . BR E A K I N G S T I M U L U S C O N T I N U I T Y
To test whether stimulus continuity is relevant for supporting long-range
oscillatory interactions in the retina, we designed a stimulus that consisted of an annular
mask superimposed on circular stimuli of various sizes (center part of the stimulus had
5°, see Figure 10). The presence of a mask outside the borders of the RFs should hinder
oscillatory responses for the center part of the stimulus, because the vicinity being not
activated should not mediate synchronization. As shown in the plots of Figure 15,
annular masks had a profound effect on oscillation amplitude, even though the rates
remained the same (rates tend to decrease with stimulus size in both unmasked and
masked conditions; N= 24, 3 different experiments). Notably, in these experiments, the
effect of stimulus size on oscillation strength seemed to reach a plateau for a stimulus of
10°.
Figure 15. Breaking stimulus continuity disrupts oscillatory responses to a circle stimulus. (Left panels) Strong oscillatory responses to a high contrast circle of 9° (upper panel) are disrupted by a dark mask cutting off the central region of the stimulus over the RF (middle panel). Oscillation frequency, 99 Hz. These effects cannot be explained by a simple decrease in luminance level because oscillations are also disrupted for a larger stimulus (12°, lower panel). Sliding window size, 200 ms; step, 100 ms. (Right plots) Population data. Notice that oscillation amplitude is always lower for annular masked stimuli. Analysis window, 1000 ms (dashed line).
0.10
3° 5° 7° 9° 10° 12° 14°
Rat
e (s
p/s)
S t imulus s ize
0.00
360
0
Tim
e (m
s)
80
0
-80
Tim
e (m
s)
80
0.3
0.0
0
-80
Co
inci
den
ces
12°
T ime3 0 0 m s
cgl03d1207 4-4
cgl03d1011 4-4
9°
Osc
. A
mpl
itud
e
Tim
e (m
s)
80
0
-80
cgl03d1207 4-4
9°
Fast osci l lat ions in the ret ina by Fabio Freitag 33
Notice, however, that although oscillation strength was much diminished for the
masked stimuli, they did not vanish completely. In the traces of Figure 15 one sees a
clear trend for an increase in oscillation amplitude (without reaching a plateau) as a
function of the size of the peripheral part of the stimulus (the center part remains
unchanged). A possible confounding factor could be the stray-light of the CRT monitor.
It would be interesting to use different types o display (e.g.., fast LCD monitor) to
dispel these interferences. Results for annular masked stimuli are given in Table 2.
4.3 . PE R C O L AT I O N S T I M U L U S
To further study whether oscillation dynamics are related to stimulus
connectedness (Neuenschwander and Singer, 1996) we used the percolation stimulus
proposed by Stephens et al. (2006) (see Methods). The sliding window analysis
presented in Figure 16 shows an example of how oscillatory behavior changes as a
function of adding pixels in the image. Although one can see a sharp appearance of an
Figure 16. Percolation threshold does not lead to a phase transition in the oscillatory responses. (Left panels) Sliding window analysis show that oscillatory responses appear much later for the percolation stimulus (see Methods) as compared to the linear ramp circular stimulus (linear change in luminance, symbolized by a ramp over a circle). Arrows indicate the expected time of percolation threshold (at 60 % of elapsed time, see Stephens et al., 2006). Notice that in this example a transition in oscillatory behavior happens just before the point in time of percolation threshold (Right plot). Population data for oscillation onset from the start of the stimulus. Although oscillatory responses took longer to appear for the percolation stimulus as compared to the linear ramp stimulus, oscillation onsets were clearly below the expected time of percolation threshold (depicted as a dashed line in the plot).
1800
Tim
e (m
s)
S t imulus protocol
0Percolation Linear
OSCILLATION ONSET
Tim
e (m
s)
0
-80
80
0.25
0.0
0
-80
Co
inci
den
ces
T ime 5 0 0 m s
cgl02e0201 1-1
cgl02e0202 1-1
80
15°
15°
Fast osci l lat ions in the ret ina by Fabio Freitag 34
oscillatory pattern at 1200 ms, at this point in time the stimulus has not yet reached
percolation threshold (1800 ms). Population data confirmed these results. In all cases
studied oscillations appeared before the expected time of percolation (mean of 640 ms
after stimulus onset). Thus, our population results do not support the predictions by
Stephens et al. (2006). In their model, a clear phase transition in oscillation should
occur just before the point of percolation. These was simply not the case in most of our
experiments.
4.4 . OS C I L L AT O RY R E S P O N S E S T O DY N A M I C A L S T I M U L I
An important goal in this study was to examine whether retinal oscillations
appear also for responses to naturalistic stimulus (Figure 17). We have used two
different gray-scale natural scene movies (200 gray values, referred as PanLab and
Medikament movies, same stimuli used in the study of Haslinger et al., 2012). These
movies were chosen because they contain scenes with relatively large uniform surfaces
(scenes of the laboratory with high spatial and temporal correlation). In addition we
built a simplified, binary version of these two films (2 luminance values, black and
white). Sliding window analysis revealed epochs of strong oscillatory responses (ON-
responses) probably coinciding with the appearance of large uniform segments or blobs
in the image (quantitative analysis is missing). Oscillations were much stronger for the
binary than for the gray-level movies. Notice that oscillatory responses were not graded,
but seemed to appear in an all-or-none fashion. This probably happened because a given
segment in the image needed to reach a certain size threshold to trigger an oscillatory
response. Interestingly, during the stimulus presentation, oscillation frequencies
remained constant across the oscillatory episodes.
Responses to flickering stimuli (random checkerboards), on the contrary, always
evoked non-oscillatory responses (Figure 17).
Fast osci l lat ions in the ret ina by Fabio Freitag 35
Figure 17. Fast oscillations appear only briefly for homogeneous image segments or blobs in natural scene movies. (Left panels) LabPan movie. Oscillation frequency for gray-level movie, 92 Hz; for binary movie, 86 Hz. (Right panels) Medikament movie. Oscillation frequency for gray-level movie, 92 Hz; for binary movie, 81 Hz. Note that oscillation frequencies are fairly constant along the trial both for the gray-level and for the binary stimulus (responses to the Medikament movie). Random flickering checkerboards evoked no oscillations. Sliding window, 200 ms; step, 100 ms.
0
cgl03d0302 4-480
-80
80
0.40
0.0
0
-80
Co
inci
den
ces
T ime 5 0 0 m s
cgl03d0305 4-4
Tim
e (m
s)
0
cgl03d0301 4-480
GRAY-LEVEL
BINARY
RANDOM
0
cgl03d0202 4-480
-80
80
0.40
0.0
0
-80
Co
inci
den
ces
T ime 5 0 0 m s
cgl03d0205 4-4
Tim
e (m
s)
0
cgl03d0201 4-480
LabPan Medikament
-80 -80
Figure 18. Oscillatory responses to gratings as function of stimulus temporal frequency (0.3 Hz, 1.25 Hz, 3.0 Hz; same spatial frequency at 1.0 cycle per degree, different velocities). (Left panels) Sliding window analysis. Oscillation frequencies were of 80 Hz, 87 Hz and 89 Hz (analysis window placed at 2000 ms). Observe the phase-reseting of the oscillatory responses for the 3.0 Hz gratings. Increase of oscillation frequency as a function of stimulus speed confirm data obtained for moving bars (Neuenschwander et al., 1999). Interestingly, the gamma oscillations in monkey V1 exhibit a similar behavior (Lima et al., 2010). Sliding window, 200 ms; step, 100 ms. (Right panels) Response histograms were computed with a resolution of 10 ms, MUA data. The transient components of the responses increase for the faster stimulus, although overall the mean firing rates are similar for all the three temporal frequencies studied.
0
cgl04c0502 5-580
-80
80
0.40
0.0
0
-80 Co
inci
den
ces
T ime5 0 0 m s
cgl04c0503 5-5
Tim
e (m
s)
0
cgl04c0501 5-580
0T ime (s )
1.0 2.0 3.0 4.0 5.0
100
0.3 Hz
1.25 Hz
3.0 Hz
cgl04c0502 5-5
cgl04c0503 5-5
cgl04c0501 5-5
Sp
ikes
/s
20°-80
Fast osci l lat ions in the ret ina by Fabio Freitag 36
Analysis of the data obtained for grating stimuli revealed that retinal oscillations
are sensitive to the temporal frequency of the stimulus (see examples in Figure 18).
Faster stimuli led to faster oscillation frequencies in the responses. Observe the
systematic decrease of oscillation frequency locked to the response onset (when the
individual bars of the gratings crossed the RF).
4.5 . DE P E N D E N C I E S O N H A L O T H A N E L E V E L S
An unexpected discovery in our study was that retinal oscillations are very
dependent on the halothane anesthesia.
Figure 19 documents this observation for responses to a large circular stimulus.
A decrease in the concentration of halothane (expired partial concentration) led to a
rapid fading of the oscillatory responses. This effects seemed to be non linear.
Decreasing the halothane level from 1.0% to 0.5% caused nearly complete vanishing of
the modulation in the correlograms. It is important to mention that before removing the
halothane from the ventilation mixture (70% N2O and 30% O2), we had supplemented
Figure 19. Halothane anesthesia greatly affects fast oscillations in the retina. ON-responses obtained for for circular light stimuli (sizes of 14° and 10°). (Left panels) In the absence of halothane (concentration of 0.0%), the oscillatory patterning in the responses completely vanished, without significant changes in rates. Sliding window, 200 ms; step, 100 ms (Right plots) Autocorrelation analysis for different epochs after turning the halothane vaporizer to 0.0 % (wash out recording session). Correlograms were normalized by the firing rates. Oscillation frequency, 75 Hz; MAS, 1.51. Analysis window, 1000 ms (indicated by a dashed line at the bottom of the sliding window plots). Notice that oscillation strength does not seem to be linearly correlated with halothane concentration. Data from the same recording sites. Halothane concentration levels were obtained from readings of the life monitoring system (expiratory gas mixture).
Tim
e (m
s)
80
0
-80
Tim
e (m
s)
80
0.3
0.0
0
-80
T ime5 0 0 m s
cgl03e0413 4-4
cgl03e0913 4-4
14°
-80 0 80Co
inci
den
ces
Co
inci
den
ces
T ime (ms)
0.8%
0.0%
1.0%
0.5%
0.0%
H A L O T H A N E
H A L O T H A N E
H A L O T H A N E
cgl03e1209 4-4
10°
75 Hz1.51
Fast osci l lat ions in the ret ina by Fabio Freitag 37
the anesthesia with ketamine (10 mg/kg body weight, i.m.) or fentanyl (0.005 mg/kg
body weight, i.v.). This was carried out to assure that the cats suffered no discomfort
(vital parameters were continuously monitored during the recording sessions). None of
these supplementary agents changed the oscillatory behavior of the responses (data not
shown).
The effects of withdraw of halothane for responses to a linear ramp stimulus are
shown in Figure 20. Lowering the concentration from 1.0% to 0.4% to 0.2% led to a
rapid decrease in oscillation modulation, despite a pronounced increase in the overall
responsiveness of the cells. Interestingly, in this case there was a clear change in the
oscillation dynamics, with the appearance of weak slow rhythmic components (about 20
Hz) at 0.2% of halothane.
Figure 20. Effects of halothane on LGN responses to a linear ramp stimulus. (Left panels) Sliding window analysis. Oscillation frequency of 72 Hz. Notice that oscillation frequency remains unchanged for the different concentration levels of halothane. Sliding window, 200 ms; step, 100 ms. (Right panels) MUA response histograms. Bin, 10 ms. In this particular example halothane anesthesia diminished substantially the overall response rates (49% decrease in firing rates). Even though populations in the retina fire less during halothane anesthesia, they oscillate strongly. Data from a different cat.
12°
0
cgl04c0502 5-580
-80
80
0.40
0.0
0
-80
Co
inci
den
ces
T ime3 0 0 m s
cgl04c0503 5-5
Tim
e (m
s)
0
cgl04c0501 5-580
0T ime (s )
1.0 3.0
100
cgl04c0502 5-5
cgl04c0503 5-5
cgl04c0501 5-5
1.0%
0.4%
H A L O T H A N E
H A L O T H A N E
0.2%H A L O T H A N E
2.0S
pik
es/s
Fast osci l lat ions in the ret ina by Fabio Freitag 38
Plots of oscillation strength as a function of halothane levels for responses to a
linear ramp stimulus are presented in Figure 21. Oscillation amplitude decreases
remarkably as a function of halothane concentration levels, despite a slight increase in
firing rates. Thus, removal of halothane lessens inhibition in overall activity, resulting in
a new non-oscillatory dynamics.
Similarly, retinogeniculate oscillatory responses to natural scene movies vanish
in absence of halothane (Figures 22 and 23). As in the data presented in Figures 19 and
21 the dependencies on the halothane concentration seemed to be non linear. At 0.6 %
halothane concentration the oscillatory components in the responses to the movie
(Medikament) were nearly absent.
Finally, we document in Figures 24 and 25 the halothane-withdrawal effects for
grating stimuli.
In all cases studied the restoring halothane concentration to the original levels
(from 0.8 to 1.2 %) brought back the oscillatory responses in the LGN.
Figure 21. Oscillation strength as a function of halothane levels for responses to a linear ramp stimulus. Population data. Two slopes were used for luminance ramping, one fast with duration of 2000 ms (gray traces) and another slower with duration of 3000 ms (red traces). (Upper plot) Firing rates increase slightly for low values of halothane. (Lower plot) Oscillation amplitude decreases remarkably as a function of halothane concentration levels. This trend does not seem to be linear. Recovery points are shown to the right (after increasing the halothane to previous levels).
0.20
0.9%
Rat
e (s
p/s)
Osc
. Am
plitu
de
0.00
45
0
HALOTHANE0.8% 0.7% 0.5% 0.4% 0.2% 0.8%
3000 ms2000 ms
12°
Fast osci l lat ions in the ret ina by Fabio Freitag 39
Figure 22. Retinal oscillations in LGN responses to natural scene movies vanish in absence of halothane. Observe that the impact on oscillation strength does not correlate linearly with concentration levels of halothane (already at 0.6% oscillations ceased almost entirely). Firing rates are slightly increased upon halothane removal, indicating that halothane causes a slight depression in the general activity of the retinogeniculate system. Oscillation frequency, 75 Hz. MAS, 1.03. Sliding window, 200 ms; step, 100 ms. Analysis window for computation of the correlograms, 500 ms. Data obtained for the Medikament movie.
0.3
0.0
T ime (ms )
cgl03e0702 5-5
1.0%
0.6%
0.2%
H A L O T H A N E
0
-80
80
Co
inc
ide
nc
es
0.1
cgl03e0702 5-5
cgl03e0602 5-5 cgl03e0602 5-5
cgl03e0802 5-5 cgl03e0802 5-5
Co
inc
ide
nc
es
0
-80
80
0
-80
80
0-80 80
T ime500 ms
Tim
e (m
s)
75 Hz1.03
75 Hz0.25
20°
Figure 23. Oscillation strength as a function of halothane levels for responses to natural scene movies (gray-level and binary stimuli). Population data. Oscillation amplitude decreases steadily as a function of halothane levels. Recovery points are shown to the right (after increasing the halothane to previous levels). Data obtained for the LabPan movie.
0.20
Rat
e (s
p/s)
Osc
. Am
plitu
de
0.00
45
0
HALOTHANE1.0% 0.6% 0.2% 0.8%
20°B I N A RY
G R AY
Fast osci l lat ions in the ret ina by Fabio Freitag 40
Figure 24. Retinal oscillations in LGN responses to grating stimuli (temporal frequency of 3.0 Hz). Notice the gradual disappearance of the oscillatory modulation in the responses as halothane concentration levels decrease. Mean average rates remain unchanged. Oscillation frequency, 89 Hz. Sliding window, 200 ms; step, 100 ms. Response histograms bin, 10 ms.
0
cgl04c1303 5-580
-80
80
0
-80
T ime5 0 0 m s
cgl04c1503 5-5
Tim
e (m
s)
0
cgl04c0603 5-580
0T ime (s )
1.0 2.0 3.0 4.0 5.0
cgl04c0502 5-5
cgl04c0503 5-5
cgl04c0501 5-5
0.8%
0.3%
H A L O T H A N E
H A L O T H A N E
0.1%H A L O T H A N E
0.40
0.0
Co
inci
den
ces
100
Sp
ikes
/s
20°
Figure 25. Oscillation strength as a function of halothane levels for responses to grating stimuli (temporal frequencies of 0.3 Hz, 1.25 Hz and 3.0 Hz). Population data. Oscillation amplitude decreases as a function of halothane levels. Recovery points are shown to the right.
0.20
0.8%
Rat
e (s
p/s)
Osc
. Am
plitu
de
0.00
45
0
HALOTHANE0.5% 0.4% 0.3% 0.2% 0.0% 0.8%
20°0.3 Hz
1.25 Hz3.0 Hz
Fast osci l lat ions in the ret ina by Fabio Freitag 41
Finally, we found evidence that isoflurane, an anesthetic agent with similar
pharmacological properties to halothane, is also capable of inducing strong oscillations
in the retinogeniculate pathways (Figure 26).
Taken together, these findings indicate that halogenated anesthetics have a
profound effect in the temporal patterning of the responses. As discussed below these
results question whether fast oscillations in the retina contribute to encode visual
information.
Figure 26. Isoflurane, an halogenated anesthetic similar to halothane, is also responsible for the appearance of fast oscillations in the retinogeniculate system. Stimulus consisted of a high contrast light circle flashed over the RFs (stimulus size, 12°). Oscillation frequency, 76 Hz. MAS, 0.38. Data was obtained after decreasing halothane level to 0.0 % and subsequently, after approximately 30 min washing period, delivering isoflurane at 1.0% (with another vaporizer).
cgl03e1611 5-5
76 Hz0.38
0
-80
80cgl03e1611 5-
1.0%I S O F L U R A N E
cgl03e0911 5-5 cgl03e0911 5-5
0
-80
80
0.0%H A L O T H A N E
0.25
0.0
T ime (ms )
Co
inc
ide
nc
es
0.1
5
Co
inc
ide
nc
es
0-80 80
T ime500 ms
Tim
e (m
s)
12°
Fast osci l lat ions in the ret ina by Fabio Freitag 42
1. DI S C U S S I O N
The primary goal of the present study was to investigate the role of fast retinal
oscillations in encoding visual information. Our study bears on the LGN for various
reasons. First, it is much simpler to record from the thalamus. To record directly from
the retina implies positioning the electrodes through the sclera, a complex and difficult
procedure. Moreover, data from the LGN will allow us to easily draw comparisons
between anesthetized and behaving cats in future experiments, as recording directly
from the eye in non-anesthetized animals is nearly impossible. Second, it is well
established that the temporal pattern of neuronal activity originated in the retina is
transmitted to the LGN (Doty and Kimura, 1963; Laufer and Verzeano, 1967; Arnett,
1975; Neuenschwander et al., 1996; Castelo-Branco et al., 1998). In these studies in the
cat, synchronization of oscillatory responses was observed only for LGN cells receiving
inputs from the same eye, independent whether the cells were located in the same LGN
or in the LGNs of the two hemispheres (Figure 6). Accordingly, simultaneous
recordings from the retina, LGN and cortex (Castelo-Branco et al., 1988) have shown
that correlated activity from the retina propagates to the primary visual cortex (areas
A17 and A18). Since the cat retina does not receive centrifugal projections and there are
no thalamic interactions across the midline we can conclude that the synchronization
mechanism of oscillatory responses in the LGN resides in the retina (see discussion in
Neuenschwander et al., 2002). Moreover, as shown recently, most LGN neurons have
one retinal ganglion cell input that accounts for nearly all LGN spikes sent to the cortex
(Sincich et al., 2007). The retina and the LGN have approximately the same number of
cells and most LGN principal neurons receive inputs from a single retinal ganglion cell.
Thus, spiking activity in the LGN faithfully follows the driving retinogeniculate inputs.
For these reasons, in this study we attribute our main conclusions to the retina,
since we know that the fast oscillations we describe here in the LGN are actually
originated in the retina.
Fast osci l lat ions in the ret ina by Fabio Freitag 43
5.1 ST I M U L U S D E P E N D E N C I E S
Since Kuffler (1953) it has been known that the size of simple stimuli modulates
the responses of a given neuron both in the retina and the LGN (Barlow et al., 1954;
Hubel and Wiesel, 1961). Small spots of light flashed over the center of the RFs evoke
strong responses. For larger stimuli reaching the surround regions of the RFs, however,
the responses diminish. Thus, response rates of single retinal ganglion cells are unlikely
to be appropriate for encoding stimulus size.
The discovery of stimulus-dependent oscillatory responses in the retina raised an
interesting alternative hypothesis. Stimulus global characteristics such as size and
continuity could be represented by a temporal code. In the cat, it has been demonstrated
that Ganzfeld illumination induces strong oscillatory responses in the retina and LGN
(Doty and Kimura, 1963; Laufer and Verzeano, 1967). Later on, Neuenschwander and
collaborators have shown that oscillation strength is correlated with stimulus size, using
light rectangles flashed over the RFs (Neuenschwander et al., 1999; Stephens et al.,
2006). Our study confirm these findings. As shown in Figure 13, only circle sizes more
than 5° degrees (approximately 3X the size of the RFs) were able to induce oscillatory
patterns in the responses. Thus, a critical mass of activated cells must be reached for the
emergence of oscillatory responses. It has been suggested that fast oscillations in the
retina arises from local interactions among retinal ganglion cells (Laufer and Verzeano,
1967; Neuenschwander et al., 1999). This may occur by the continuous phase-locking
across neighboring cells over large extensions in the retinal mosaic (see example of
correlated activity in Figure 5).
To test this hypothesis, we used an annular mask to break continuity in space
(Figure 14). We expected to see no oscillations for the cells responding to the central
part of the stimulus (stimuli smaller than 5 degrees induced only weak oscillations, see
also Figure 13), as they should be functionally disconnected from the large interacting
populations responding to the peripheral part. The population data presented in Figure
14 show, however, that oscillations may persist for annular stimuli with large surrounds,
although in general there was a significant decrease in the oscillatory behavior of the
responses for the masked stimuli as compared to the controls. As mentioned before, a
Fast osci l lat ions in the ret ina by Fabio Freitag 44
possible confounding factor is the stray-light coming from the monitor, and controls
should be made to verify this problem.
Nevertheless, it is possible to conceive that synchronization of oscillatory
responses in the retina have a different mechanism than local interactions of contiguous
cells. Amacrine cells with large dendritic arbors are a likely candidate to mediate these
interactions. In principle, they could functionally bridge regions across considerable
distances in the retinal surface, working as hubs in a small-world network (Figure 27).
Besides the masked circular stimuli, we used the percolation stimulus proposed
by Stephens et al. (2006) to test whether continuity in space is critical for generation of
coherent oscillations in the retina. The central idea here is that a phase transition exists
when local interactions among neighboring cells entrain the whole network to a global
synchronous dynamics. This point should correspond to the percolation threshold of the
stimulus (60% of the stimulus duration). A this point the clusters of light pixels in the
image are randomly connected throughout the image (connectedness is independent of
shape). The population data presented in Figure 16, however, do not seem to support
this model since the onset of oscillatory responses does not coincided with the expected
timing of percolation threshold. A confounding factor in these experiments is the
increase of global luminance as more pixels are added. New experiments are needed to
exclude these problems.
Figure 27. Example of a small-world network (National Institute of Organization Dynamics in Australia). A similar organization in the retina would explain saltatory long-range horizontal interactions. Amacrine cells with their large dendritic fields could work as hubs in the network.
Fast osci l lat ions in the ret ina by Fabio Freitag 45
Finally, we have applied a variety of complex stimuli to follow the dynamics of
the retinal oscillatory responses in more naturalistic conditions. As shown in the Figure
17, oscillations in response to natural scene stimuli appeared only for short epochs,
likely when large connected segments in the image (e.g., surfaces or the faces of a
geometrical object) covered the RFs. Accordingly, oscillations were stronger for the
binary images, which had larger uniform segments than the gray-scale images. It would
be interesting to match the precise position of the RFs over the image sequences in
order to follow the oscillatory behavior as function of blob size. Qualitatively, we can
see that the critical blob size for triggering oscillations is very large. In this sense,
encoding of size would be in place only after a certain threshold is exceeded.
An open issue in this study is how the timing of the stimulus affects the retinal
oscillations. We could observe that oscillatory responses vanished for dynamical
flickering stimuli (Figure 17). Moreover, in additional experiments with large flickering
stimuli of various temporal frequencies we could verify that oscillatory responses are
disrupted for stimuli faster than 20 Hz (data not shown). Thus, oscillatory responses in
the retina are in general limited to relatively large and slow stimuli. From a theoretical
perspective, this would represent a severe limitation for the encoding bandwidth of the
system.
At the present we do not know whether retinal oscillations are meaningful for
cortical processing. The precise timing of synchronous spikes in the LGN are important
for effectively driving cortical targets (Alonso et al., 1996; Stanley et al., 2012). It is
possible that oscillatory inputs to the cortical neurons convey information about
-π 0 π
Figure 28. Visual processing using coupled oscillators (modified from Koepsell et al., 2010). (Left) Sample image (from the Berkeley segmentation database). (Middle) Phase representation of the image in a network of coupled phase oscillators with oscillation frequency, coupling parameters depending on similarity of pixel luminance value (model proposed by Koepsell et al., 2010). (Right) Information represented by cells spiking at a specific oscillation phase. Phase-synchronization of oscillatory responses may encode image contours relevant for further cortical processing.
Fast osci l lat ions in the ret ina by Fabio Freitag 46
homogeneous image segments and contribute to scene segmentation, as hypothesized in
a few theoretical studies (Stephens et al., 2006; Koepsell et al., 2010; Figure 28). It will
be important to further extend these conclusions, which were drawn from experiments
in the anesthetized and paralyzed cat, to the behaving monkey.
5.2 EF F E C T S O F H A L O T H A N E A N E S T H E S I A
We found abundant evidence, with a variety of visual stimuli, that retinal
oscillations in the cat are dependent on halothane anesthesia (Figures 19 to 24). Several
seminal studies on temporal coding in the retinogeniculate system have used
anesthetized animals as experimental model, applying as anesthetic agent either sodium
thiopental (Doty and Kimura, 1963; Laufer and Verzeano, 1967; Reinagel and Reid,
2000; Butts et al., 2007; Desbordes et al., 2008), halothane (Neuenschwander et al.,
1996; Castelo-Branco et al., 1998) or isoflurane (Ito et al., 2010). Halogenated
anesthetics, such as halothane, are known to act directly on GABAA receptors through
an agonist effect by prolonging the decay of inhibitory postsynaptic currents and
increasing IPSP amplitudes (Li et al., 2000; Nishikawa and Maclver, 2000). Halothane
is known to enhance receptor affinity (Krasowski and Harrison, 1999) and also inhibits
GABA uptake (Kotani and Akaike, 2013). Generally, inhibition mediated by GABAA
receptors plays an important role in the generation of high frequency oscillations in the
hippocampal (Wang and Buzsáki, 1996; see review in Buzsáki and Wang, 2012) and the
cortical (Hasenstaub et al., 2005) networks. Similarly, in the thalamus, inhibitory cells
are key players in providing inhibition to thalamocortical cells and therefore are directly
involved in the generation of thalamic oscillations (Steriade et al., 1996). Halothane was
also shown to facilitate evoked gamma in the rat visual cortex (Imas et al., 2004).
Knowing that GABAA receptors are abundant in the outer plexiform layer
(OPL) of the retina (Hughes et al., 1991 ; Vardi et al., 1992), it is natural to think that
these agents interfere with the delicate balance of excitation/inhibition in the networks.
In fact, in the isolated retina of the frog, applying bicuculline (a GABAA-receptor
antagonist) suppresses synchronous oscillations of the dimming detectors (Ishikane et
al, 1999; Arai et al., 2004). However, the mechanism for the generation of the fast
oscillations in the retina is still largely unknown. Why augmenting inhibition leads to
Fast osci l lat ions in the ret ina by Fabio Freitag 47
strong oscillations in the retinal circuits? Are the retinal oscillations generated in the
outer plexiform layer (OPL)? These are important questions that should be addressed by
in-vivo pharmacological studies.
Even if halothane is not directly responsible for the retinal oscillation it and
other anesthetics like it clearly have a facilitatory effect, which must be factored in to
any model using data derived using an anesthetized preparation. Interestingly, the
dependency of oscillation amplitude on halothane concentration seemed to be non-
linear. For example in Figure 19 oscillatory responses are already absent for a
concentration level of 0.5%. Similarly, data from a different cat show that at a
concentration of 0.6% the oscillations are nearly vanished for responses to natural scene
stimuli (Figure 22). Possibly a phase transition point must be reached for the emergence
of oscillatory pattens in the retinal networks. Importantly, the oscillatory dynamics
cannot be explained by firing rates (ringing of maximal responses). Actually, in our
experiments, halothane generally caused a slight decrease in response rates.
In this study, we show that isoflurane also evoked oscillatory responses with the
same temporal characteristics as halothane. Thus, our findings can be generalized to
halogenated anesthetics. This is important since these two agents have been routinely
used in many studies of temporal coding at various levels in the sensory systems (e.g.,
Neuenschwander and Singer, 1996; Ito et al. 2010).
Finally, a few studies have reported that retinal oscillations occur also for non-
anesthetized and paralyzed cats and monkeys (after transpontine transection) (Doty and
Kimura, 1963), which is at odds with our results. In these early experiments, however,
the stimulus conditions were very different from ours. Visual stimulation was generally
made by illuminating the whole visual field using light flashes (Doty and Kimura, 1963;
Laufer and Verzeano, 1967). It remains to be investigated whether halothane is
responsible to the generation of oscillations in the retina, or if it simply amplifies them.
These unresolved issues should be clarified by recordings in awake behaving animals.
Fast osci l lat ions in the ret ina by Fabio Freitag 48
2. CO N C L U S I O N
In this study we recorded MUA responses in the LGN in order to investigate the
role of fast retinal oscillations in visual encoding. In accord with previous studies,
stimulus size and luminance have a strong effect on the modulation of oscillation
strength. Moreover, oscillations are disrupted by changes in stimulus uniformity in
space and time. Images made of connected clusters of pixels with random shapes are
capable of inducing strong oscillatory responses. In our experiments, however, we were
not able to demonstrate the appearance of a phase transition in the oscillatory behavior
of the responses at stimulus percolation threshold, as predicted by Stephens et al., 2006.
Stimuli with complex temporal and spatial dynamics, such as natural scene movies, are
also able to elicit oscillatory responses, but only for short epochs. Overall, these results
support the idea that retinal oscillations serve for encoding of large, continuous and
uniform segments in the visual image.
A surprising finding in this study was that retinal oscillations are highly
dependent on halothane anesthesia levels. In the absence of halothane, oscillatory
activity was rarely induced by the visual stimuli even when they were optimal. The
same findings were obtained for isoflurane, which has similar pharmacological
properties. Taking into consideration these new and unexpected results we question
whether feedfoward oscillations in the early visual system are simply due to an
imbalance between excitation and inhibition in the retinal circuits created by the use of
halogenated anesthetics. Are retinal oscillations (at least in the cat) an artifact from the
halothane anesthesia? Further studies in awake behaving animals are necessary to
extend these conclusions.
Fast osci l lat ions in the ret ina by Fabio Freitag 49
3. RE F E R E N C E S
Ahissar, E., Sosnik, R., Haidarliu, S. (2000). Transformation from temporal to rate coding in a somatosensory thalamocortical pathway. Nature. 406, 302 – 306;
Alonso, J.M., Usrey, W.M., Reid, R.C. (1996). Precisely correlated firing in cells of the lateral geniculate. Nature. 383 (1996) 815–819;
Andolina, I.M., Jones, H.E., Wang, W., Sillito, A.M. (2007). Corticothalamic feedback enhances stimulus response precision in the visual system. PNAS. 104, 1685 – 1690;
Arai, I., Yamada, Y., Asaka, T., Tachibana, M. (2004). Light-Evoked Oscillatory Discharges in Retinal Ganglion Cells Are Generated by Rhythmic Synaptic Inputs. Journal of Neurophysiology. 92, 715 – 725;
Ariel, M., Daw, N.W., Rader, R.K. (1983). Rhythmicity in rabbit retinal ganglion cell responses. Vision Research. 23, 1485 – 1493;
Arnett, D.W. (1975). Correlation analysis of units recorded in the cat dorsal lateral geniculate nucleus. Experimental Brain Research. 24 (1975) 111–130;
Barlow, H. B., Fitzhugh, R., Kuffler, S.W. (1954). Resting discharge and dark adaptation in the cat, Journal of Physiology. (Lond.), 25 28–9P;
Bishop, G.H. (1933). Fiber groups in the optic nerve. American Journal of Physiology. 106, 460 – 470;
Bishop, P.O., Levick, W.R., Williams, W.O. (1964). Statistical Analysis of the Dark Discharge of Lateral Geniculate Neurones. Journal of Physiology. 170: 598 - 612;
Butts, D.A., Weng, C., Jin, J., Yeh, C., Lesica, N.A., Alonso, J.M., Stanley, G.B. (2007). Temporal precision in the neural code and the timescales of natural vision. Nature. 449, 92 – 96;
Buzsáki, G. and Wang, X.J. (2012). Mechanisms of Gamma Oscillations. Annual Review of Neuroscience. 35, 203 – 25;
Casagrande, V.A. and Kaas, J.H. (1994). The afferent, intrinsic, and efferent connections of primary visual cortex in primates. Cerebral Cortex. 10, 201 – 259;
Castelo-Branco, M., Goebel, R., Neuenschwander, S., Singer, W. (2000). Neural synchrony correlates with surface segregation rules. Nature. 405, 685 – 689;
Castelo-Branco, M., Neuenschwander, S., Singer, W. (1998). Synchronization of visual responses between the cortex, lateral geniculate nucleus, and retina in the anesthetized cat. The Journal of Neuroscience. 18, 6395 – 6410;
Dan, Y., Alonso, J.M., Usrey, W.M., Reid, R.C. (1998). Coding of visual information by precisely correlated spikes in the lateral geniculate nucleus, Nature Neuroscience. 1, 501–507;
Desbordes, G., Jin, J., Weng, C., Lesica, N.A., Stanley, G.B., Alonso, J.M. (2008). Timing Precision in Population Coding of Natural Scenes in the Early Visual System. Plos Biology. 6, 2672 – 2682;
Fast osci l lat ions in the ret ina by Fabio Freitag 50
Doty, R.W., and Kimura, D.S. (1963). Oscillatory Potentials in the Visual System of Cats and Monkeys. Journal of Physiology. 168, 205 – 218;
Eckhorn, R., Bauer, R., Jordan, W., Brosch, M., Kruse, W., Munk, M., Reitboeck, H.J. (1988). Coherent Oscillations: A Mechanism of Feature Linking in the Visual Cortex? Biological Cybernetics. 60, 121 – 130;
Engel, A.K., Fries, P., Singer, W. (2001). Dynamic predictions: oscillations and synchronization in top-down processing. Nature. 2, 704 – 716;
Engel, A.K., König, P. and Singer, W. (1991). Direct physiological evidence for scene segmentation by temporal coding. PNAS. USA 88, 9136 – 9140;
Enroth-Cugell, C., and Robson, J.G. (1966). The contrast sensitivity of retinal ganglion cells of the cat. Journal of Physiology. 187, 517 – 552;
Fries, P., Reynolds, J.H., Rorie, A.E., Desimone, R. (2001). Modulation of Oscillatory Neuronal Synchronization by Selective Visual Attention. Science. 291, 1560 – 1563;
Fries, P. (2009). Neuronal gamma-band synchronization as a fundamental process in cortical computation, Annual Review of Neuroscience. 32, 209–224;
Fröhlich, F. W. (1914). Beiträge zur allgemeinen Physiologie der Sinnesorgane. Psychologische Physiologie Sinnesorganische II. Abteilung Sinnesphysiologische, 48, 28–164;
Gasser, H.S. and Erlanger, J. (1929). Role of fiber size in establishment of nerve block by pressure or cocaine. American Journal of Physiology. 88, 581–591;
Gattass, R., Nascimento-Silva, S., Soares, J.G.M., Lima, B., Jansen, A.K., Diogo, A.C.M., Farias, M.F., Marcondes, M., Botelho, E.P., Mariani, O.S., Azzi, J., Fiorani, M. (2005). Cortical visual areas in monkeys: location, topography, connections, columns, plasticity and cortical dynamics. Philosophical Transactions of the Royal Society B. 360, 709 –731;
Gollisch, T. and Meister, M. (2008). Rapid Neural Coding in the Retina with Relative Spike Latencies. Science. 319: 1108 - 1111;
Gotch F. (1903). The time relations of the photo-electric changes in the eyeball of the frog. Journal of Physiology. 29(4-5):388–410;
Gray, C.M. and Viana Di Prisco, G. (1997). Stimulus-dependent neuronal oscillations and local synchronization in striate cortex of the alert cat. The Journal of Neuroscience. 17, 3239–3253;
Gray, C.M. and Singer, W. (1989). Stimulus-specific neuronal oscillations in orientation columns of cat visual cortex. PNAS. USA. 86, 1698 – 1702;
Gray, C. M., König P., Engel A. K., Singer W. (1989). Oscillatory responses in cat visual cortex exhibit inter-columnar synchronization which reflects global stimulus properties, Nature. 338, 334–337;
Granit, R., Therman, P.O. (1935). Excitation and inhibition in the retina and in the optic nerve, Journal of Physiology. (Lond.). 83, 359–381;
Fast osci l lat ions in the ret ina by Fabio Freitag 51
Hasenstaub, A., Shu, Y., Haider, B., Kraushaar, U., Duque, A., McCormick. (2005). Inhibitory Postsynaptic Potentials Carry Synchronized Frequency Information in Active Cortical Networks. Neuron. 47, 423 – 435;
Haslinger, R., Pipa, G., Lima, B., Singer, W., Brown, E.N., Neuenschwander, S. (2012). Context Matters: The Illusive Simplicity of Macaque V1 PLoS ONE. 7 (2012) e39699;
He, D.S., Burt, J.M. (2000). Mechanism and Selectivity of the Effects of Halothane on Gap Junction Channel Function. Circulation Research. 86, 104 – 109;
Herculano-Houzel, S., Munk, M. H., Neuenschwander, S., Singer, W. (1999) Precisely synchronized oscillatory firing patterns require electroencephalographic activation. The Journal of Neuroscience. 19, 3992–4010;
Hoffmann, K.P., Stone, J., Sherman, M.S. (1972). Relay of Receptive-Field Properties in Dorsal Lateral Geniculate Nucleus of the Cat. Journal of Neurophysiology. 35, 518 – 521;
Hormuzdi, S.G., Filippov, M.A., Mitropoulou, G., Monyer, H., Bruzzone, R. (2004). Electrical synapses: a dynamic signaling system that shapes the activity of neuronal networks. Biochimica et Biophysica Acta. 1662, 113 – 137;
Hubel, D.H., Wiesel, T.N. (1961). Integrative action in the cat’s lateral geniculate body. Journal of Physiology. 155, 385 – 398;
Hubel, D.H., Wiesel, T.N. (1962). Receptive fields, binocular interaction and functional architecture in the cat’s visual cortex. Journal of Physiology. 160, 106 – 154;
Hughes, T.E., Grünert, U., Karten, H.J. (1991). Receptors in the retina of the cat: an immunohistochemical study of wholemounts, sections, and dissociated cells. Visual Neuroscience. 6, 229 – 238;
Imas, O.A., Ropella, K.M., Wood, J.D., Hudetz, A.G. (2004). Halothane augments event-related γ oscillations in rat visual cortex. Neuroscience. 123, 269 – 278;
Ishikane, H., Gangi, M., Honda, S., Tachibana, M. (2005). Synchronized retinal oscillations encode essential information for escape behavior in frogs. Nature Neuroscience. 8, 1087 – 1095;
Ishikane, H., Kawana, A., Tachibana, M. (1999). Short- and long-range synchronous activities in dimming detectors of the frog retina. Visual Neuroscience. 16, 1001 – 1014;
Ito, H., Maldonado, P.E., Gray, C.M. (2010). Dynamics of Stimulus-Evoked Timing Correlations in the Cat Lateral Geniculate Nucleus. Journal of Neurophysiology. 104, 3276 – 3292;
Kirschfeld, K. (1992). Oscillations in the insect brain: do they correspond to the cortical gamma-waves of vertebrates? PNAS. USA. 15, 4764 – 4768;
Koepsell, K., Wang, X., Vaingankar, V., Wei, Y., Wang, Q., Rathbun, D.L., Usrey, W.M., Hirsch, J.A., Sommer, F. (2009). Retinal oscillations carry visual information to cortex. Frontiers in systems neuroscience. 3, 1 – 18;
Koepsell, K., Wang, X., Hirsch, J.A., Sommer, F.T. (2010). Exploring the function of neural oscillations in early sensory systems. Frontiers in Neuroscience. 4, 53 – 61;
König, P. (1994) A method for the quantification of synchrony and oscillatory properties of neuronal activity. The Journal of Neuroscience Methods. 54:31–37;
Fast osci l lat ions in the ret ina by Fabio Freitag 52
Kotani, N. and Akaike, N. (2013). The effects of volatile anesthetics on synaptic and extrasynaptic GABA-induced neurotransmission. Brain Research Bulletin. 93, 69 – 79;
Kreiter, A.K. and Singer W. (1996). Stimulus-dependent synchronization of neuronal responses in the visual cortex of the awake macaque monkey. The Journal of Neuroscience. 16 2381–2396;
Krasowski, M.D. and Harrison, N.L. (1999). General anaesthetic actions on ligand-gated ion channels. Cell Molecular Life Sciences. 55, 1278 – 1303;
Kratz, K.E., Webb, S.V., Sherman, S.M. (1978). Electrophysiological classification of X- and Y- cells in the cat’s lateral geniculate nucleus. Vision Research. 18, 489 – 492;
Kuffler, S. (1953). Discharge patterns and functional organization of mammalian retina. Journal of Neurophysiology. 16 37–68;
Laufer, M. and Verzeano, M. (1967). Periodic activity in the visual system of the cat. Vision Research. 7, 215 – 229;
Laurent, G. (2002). Olfactory networks dynamics and the coding of multidimensional signals. Nature. 3, 884 – 895;
Li, X., Czajkowski, C., Pearce, R.A. (2000). Rapid and Direct Modulation of Receptors by Halothane. Anestheology. 92, 1366 – 1375;
Lima, B., Singer, W., Chen, N., Neuenschwander, S. (2009). Synchronization Dynamics in Response to Plaid Stimuli in Monkey V1. Cerebral Cortex. 20, 1556 – 1573;
Lima, B., Singer, W., Neuenschwander, S. (2011). Gamma Responses Correlate with Temporal Expectation in Monkey Primary Visual Cortex. The Journal of Neuroscience. 31, 15919 – 15931;
Livingstone, M. and Hubel, D. (1988). Segregation of Form, Color, Movement, and Depth: Anatomy, Physiology, and Perception. Science. 240, 740 – 749;
Merker, B. (2013). Activation, not cognition, is the functional key to cortical gamma oscillations. Neuroscience and Biobehavioral Reviews. 37, 401 – 417;
Munk, M. H., Roelfsema, P. R., König, P., Engel, A. K., Singer, W. (1996). Role of reticular activation in the modulation of intracortical synchronization. Science. 272, 271–274;
Nassi, J.J., Callaway, E.M. (2009). Parallel Processing Strategies of the Primate Visual System. Nature Review Neuroscience. 10, 360 – 372;
Neuenschwander, S., Castelo-Branco, M., Baron, J., Singer, W. (2002). Feed-forward synchronization: propagation of temporal patterns along the retinothalamocortical pathway. Philosophical Transactions of Royal Society B. 357, 1869 – 1876;
Neuenschwander, S., Castelo-Branco, M., Singer, W. (1999). Synchronous oscillations in the cat retina. Vision Research. 39, 2485 – 2497;
Neuenschwander, S., Singer, W. (1996). Long-range synchronization of oscillatory light responses in the cat retina and lateral geniculate nucleus. Nature. 379, 728 – 733;
Fast osci l lat ions in the ret ina by Fabio Freitag 53
Nishikawa, K. and Maclver, M.B. (2000). Membrane and Synaptic Actions of Halothane on Rat Hippocampal Pyramidal Neurons and Inhibitory Interneurons. The Journal of Neuroscience. 20, 5915 – 5923;
Palanca, B.J.A. and DeAngelis, B.J.A. (2005). Does neuronal synchrony underlie visual feature grouping? Neuron. 46, 333–346;
Pipa, G., Chen, Z., Neuenschwander, S., Lima, B., Brown, E.N. (2012). Mapping of Visual Receptive Fields by Tomographic Reconstruction. Neural Computation. 24, 2543–2578;
Press. W.H., Teukolsky, S.A., Vetterling, W.T., Flannery, B.P. (1986). Numerical Recipes in C – The Art of Scientific Computing. Cambridge University Press;
Quiroga, R.Q., Nadasdy, Z., Ben-Shaul, Y. (2004). Unsupervised spike detection and sorting with wavelets and superparamagnetic clustering. Neural Computation. 16, 1661 – 1687;
Reich, D.S., Victor, J.D., Knight, B.W., Ozaki, T., Kaplan, W. (1997). Response Variability and Timing Precision of Neuronal Spike Trains in Vivo. Journal of Neurophysiology. 77, 2836 – 2841;
Reinagel, P. and Reid, C. (2000). Temporal Coding of Visual Information in the Thalamus. The Journal of Neuroscience. 20, 5392 – 5400;
Roberts, W.M. and Rutherford, M.A. (2008). Linear and nonlinear processing in hair cells. The Journal of Experimental Biology. 211, 1775 – 1780;
Roelfsema, P., Lamme, V.A.F., Spekreijse, H. (2004). Synchrony and covariation of firing rates in the primary visual cortex during contour grouping. Nature Neuroscience. 7, 982 – 991;
Rosa, M.G.P. and Tweedale, R. (2005). Brain maps, great and small: lessons from comparative studies of primate visual cortical organization. Philosophical Transactions of Royal Society B. 360, 665 – 691;
Saito, H.A. (2004). Morphology of physiologically identified X-, Y-, and W-type retinal ganglion cells of the cat. Journal of Comparative Neurology. 221, 279 – 288;
Sherman, S.M. and Guillery, R.W. (2006). Exploring the thalamus and its role in cortical function. Sec. Edition. The MIT Press. Cambridge, Massachusetts. London, England;
Sherman, S.M., Guillery, R.W. (2002). The role of the thalamus in the flow of information to the cortex. Philosophical Transactions of Royal Society B. 357, 1695 – 1708;
Sillito, A.M., Cudeiro, J., Jones, H.E. (2006). Always returning: feedback and sensory processing in visual córtex and thalamus. Trends in Neurosciences. 29, 307 – 316;
Sincich, L.C., Adams, D.L., Economides, J.R., Horton J.C. (2007). Transmission of spike trains at the retinogeniculate synapse. The Journal of Neuroscience. 27, 2683–2692;
Singer, W. (1999). Neuronal Synchrony: A versatile Code for the Definition of Relations? Neuron. 24, 49-65;
Söhl, G., Maxeiner, S., Willecke, K. (2005). Expression and functions of neuronal gap junctions. Nature. 6, 191 – 200;
Fast osci l lat ions in the ret ina by Fabio Freitag 54
Stephens, G.J., Neuenschwander, S., George, J.S., Singer, W., Kenyon, G.T. (2006). See globally, spike locally: oscillations in a retinal model encode large visual features. Biological Cybernetics. 95, 327 – 348;
Steriade, M., Contreras, D., Amzica, F., Timofeev, I. (1996). Synchronization of Fast (30 – 40 Hz) Spontaneous Oscillations in Intrathalamic and Thalamocortical Networks. The Journal of Neuroscience. 16, 2788 – 2808;
Stanley, G.B., Jin, J., Wang, Y., Desbordes, G., Wang, Q., Black, M.J. , Alonso, J-M (2012). Visual orientation and directional selectivity through thalamic synchrony. The Journal of Neuroscience, 32 (2012) 9073–9088.
Stone J. (1983). Parallel processing in the Visual System. New York: Plenum Press;Thiele, A. and Stoner, G. (2003). Neuronal synchrony does not correlate with motion
coherence in cortical area MT. Nature. 421, 366 – 370;Tolhurst, D.J. (1973). Separate channels for the analysis of the shape and the movement
of a moving visual stimulus. Journal of Physiology. 231, 385 – 402;Uhlhaas, P.J., Pipa, G., Neuenschwander, S., Wibral, M., Singer, W. (2011). A new look at
gamma? High- (>60 Hz) γ-band activity in cortical networks: Function, mechanisms and impairment. Progress in Biophysics and Molecular Biology. 105, 14-28;
Usrey, M.W. (2002). Spike timing and visual processing in the retinogeniculocortical pathway. Philosophical Transactions of Royal Society B. 357, 1729 – 1737;
Usrey, W.M., Reppas, J.B., Reid, R.C. (1998). Paired-spike interactions and synaptic efficacy of retinal inputs to the thalamus. Nature. 395, 384 – 387;
Vardi, N., Masarachia, P., Sterling, P. (1992). Immunoreactivity to Receptor in the Outer Plexiform Layer of the Cat Retina. The Journal of Comparative Neurology. 320, 394 – 397;
Von der Malsburg, C. (1981). The Correlation Theory of Brain Function. Internal Report 81-2, Dept. of Neurobiology, Max-Planck Institute for Biophysical Chemistry, Göttingern, Germany;
Wang, X. and Buzsáki, G. (1996). Gamma Oscillation by Synaptic Inhibition in a Hippocampal Interneuronal Network Model. The Journal of Neuroscience. 16, 6402 – 6413;
Wentlandt, K., Samoilova, M., Carlen, P.L., El Beheiry, H. (2006). General Anesthetics Inhibit Gap Junction Communication in Cultured Organotypic Hippocampal Slices. Anesthesia and Analgesia. 102, 1692 – 1698;
Womelsdorf, T., Schoffelen, J.M., Oostenveld, R., Singer, W., Desimone, R., Engel, A., Fries, P. (2007). Modulation of Neuronal Interactions Through Neuronal Synchronization. Science. 316, 1609 – 1612;
Wunderle, T., Eriksson, D., Schmidt, K.E. (2013). Multiplicative Mechanism of Lateral Interactions Revealed by Controlling Interhemispheric Input. Cerebral Cortex. 23, 900 – 912;
Yeh, C., Stoelzel, C.R., Alonso, J.M. (2003). Two different types of Y cells in the cat Lateral Geniculate Nucleus. Journal of Neurophysiology. 90, 1852 – 1864.
Fast osci l lat ions in the ret ina by Fabio Freitag 55
4. TA B L E S
Table 1 . Effects of stimulus size on oscillation amplitude of LGN responses. Means and standard deviations for oscillation amplitude and firing rate for circles and its related annular mask stimuli for 5 different sizes and the first two circles.
Size (degrees) Mean SD Mean SDHigh luminance Osc. Amplitude Osc. Amplitude Rate Rate
2° 0,0025 0,0071 114,91 80,863° 0,0025 0,0071 150,85 100,974° 0,0213 0,0196 174,47 103,016° 0,0350 0,0273 179,37 104,657° 0,0263 0,0297 184,24 100,548° 0,0388 0,0340 175,71 95,0510° 0,0588 0,0309 172,91 94,6711° 0,0788 0,0432 167,79 87,6612° 0,1075 0,0528 171,87 87,513° 0,2300 0,2986 176,12 90,39
Medium luminance2° 0,0038 0,0106 90,81 74,643° 0,0025 0,0071 132,26 93,224° 0,0088 0,0181 160,81 103,026° 0,0188 0,0210 168,60 100,947° 0,0325 0,0212 170,81 98,098° 0,0400 0,0262 169,06 96,1810° 0,0375 0,0328 167,85 94,6811° 0,0538 0,0333 170,72 87,9312° 0,0725 0,0373 167,99 88,3513° 0,1063 0,0478 173,56 93,01
Low luminance2° 0,0000 0,0000 61,50 38,663° 0,0025 0,0071 93,01 55,744° 0,0013 0,0035 118,06 64,056° 0,0113 0,0181 127,89 70,077° 0,0050 0,0141 127,60 71,898° 0,0100 0,0185 132,40 77,4510° 0,0100 0,0214 134,59 80,2211° 0,0338 0,0292 130,34 81,4312° 0,0488 0,0295 130,5 78,5113° 0,0638 0,0478 127,16 79,05
Fast osci l lat ions in the ret ina by Fabio Freitag 56
Table 2 . Results for the annular mask protocol. Comparisons between masked and unmasked stimuli (circle). Mean values and standard deviation for oscillation amplitude and firing rate of the 10 different sizes at 3 different levels of luminance.
Protocol Mean SD Mean SDStimulus size Osc. Amplitude Osc. Amplitude Rate Rate
3° - Circle 0,0050 0,0117 212,97 79,465° - Circle 0,0167 0,0235 212,83 80,837 °- Annular mask 0,0092 0,0138 200,29 75,217 ° - Circle 0,0467 0,0323 190,47 78,529 ° - Annular mask 0,0133 0,0156 177,00 70,169 ° - Circle 0,0608 0,0425 169,52 71,7010 ° - Annular mask 0,0292 0,0300 159,10 69,9410 ° - Circle 0,0767 0,0352 153,65 67,8612 ° - Annular mask 0,0383 0,0346 143,84 63,9612 ° - Circle 0,0758 0,0408 142,79 65,6214 ° - Annular mask 0,0533 0,0274 141,75 63,0614 °- Circle 0,0758 0,0337 134,48 60,94
Fast osci l lat ions in the ret ina by Fabio Freitag 57
5. AC K N O W L E D G M E N T S
Many thanks to my advisor Sergio Neuenschwander, who is a special scientist
and person with skills in many fields. A “key figure” on my formation as a scientist
during this process and an inspiring reference. I learned with him the beauty of making
figures for a manuscript.
I also wish to thank Kerstin E. Schmidt for gently providing the space and all
conditions for the experiments performed in this study.
Thanks to Edward J. Tehovnik for valuable discussions about science, and
Sergio Conde and Luiz Lana for data analysis support. Special thanks to Jailson, Monik
and Wittyla for animal care and assistance in the laboratory. I would like specially to
thank my colleagues in the Instituto do Cérebro - UFRN, where I had the opportunity to
learn a lot from professors and students in a very warm atmosphere. Thanks to Maria
Salete B. Freitag for her continuous support and assistance, and for being such a strong
personal reference, and also to my close friends for support.
Many thanks to Prof. Jerome Baron and Prof. Adriano Tort for kindly accepting
to evaluate my thesis work, and Akaline Araújo for her secretary work at the PG-Neuro,
UFRN.
During this work I was supported by CAPES (Master‘s Degree Scholarship),
and the laboratory of Sergio Neuenschwander was supported by CNPq and a
collaboration between the Brain Institute-UFRN and the Max-Planck Institute for Brain
Research, Frankfurt.
Fast osci l lat ions in the ret ina by Fabio Freitag 58