Exp Brain Res (2009) 194:339–346

DOI 10.1007/s00221-009-1703-y

RESEARCH ARTICLE

Non-visually evoked activity of isthmo-optic neurons in awake, head-unrestrained quail

Hiroshi Ohno · Hiroyuki Uchiyama

Received: 4 December 2008 / Accepted: 4 January 2009 / Published online: 28 January 2009© Springer-Verlag 2009

Abstract Changes in the internal state of the brain maymodulate retinal function. In birds, most neurons in theisthmo-optic (IO) nucleus project their axons topographi-cally into the contralateral retina, and activity in IO neuronsenhances visual responses of retinal ganglion cells in thetarget retinal region. To elucidate the signiWcance of thispathway, we recorded spikes of IO neurons in four awakeJapanese quail using an implanted electrode assemblywhile recording unrestrained head movements. The IO neu-rons Wred passively in response to visual stimuli in recep-tive Welds and non-visually without visual stimuli or eye–head movements. Non-visually evoked activity wasobserved in the middle of eye–head Wxation, as well as atabout 200 ms before the onset of head saccades. Intensityof activity before onset of head saccades depended on thedirection of motion of subsequent head saccades. Local ret-inal output may be enhanced by centrifugal signals beforegaze shifts.

Keywords Retina · Isthmo-optic nucleus · Tectum · Centrifugal pathway · Saccade · Attention

Introduction

In 1888, Ramón y Cajal found axons projecting from anextraretinal source to the avian retina (Ramón y Cajal1889). Since then, the potential for centrifugal control ofretinal function has attracted the interest of many investiga-

tors (for historical reviews, see Cowan 1970; Uchiyama1989; Uchiyama and Stell 2005). Over the last Wve decades,several distinctive properties of the avian retinopetal sys-tem, the system projecting centrifugally to the retina, havebeen elucidated. The isthmo-optic nucleus (ION) was con-Wrmed as the source of retinopetal projections to the avianretina (Cowan 1970). Most neurons in the ION project theiraxons into the contralateral retina, making a contact withsingle isthmo-optic (IO) target cells (Dowling and Cowan1966; Uchiyama et al. 2004, 1995; Uchiyama and Ito1993). In turn, the IO neurons receive input from neurons inthe ipsilateral optic tectum (McGill et al. 1966; Wallenberg1898), and the output of these tecto-IO neurons may berestricted to single IO neurons (Uchiyama and Watanabe1985; Uchiyama et al. 1996). The avian centrifugal visualsystem is thus composed of discrete, parallel modules con-sisting of three serially connected neurons (a tecto-IO neu-ron, an IO neuron, and an IO target cell) (Fig. 1).

Physiological and theoretical studies have indicated thatretinopetal modules globally compete for activity, evenamong modules whose receptive Welds in the visual Weldare distant from each other, possibly allowing fewermodules to be simultaneously active (Uchiyama 1999;Uchiyama et al. 1998). Electrical stimulation of the IONtransiently enhances the visual response of the retinal gan-glion cells (Miles 1972b; Uchiyama and Barlow 1994), andthe single retinopetal modules aVect the activity in a localretinal area, corresponding to a visual angle of 1°–3°(Uchiyama et al. 2004). The retinopetal system may thusfunction as a mechanism for local, transient gain enhance-ment of retinal output. The retinopetal system is alsoknown to receive top-down excitatory signals from anavian homolog of the mammalian visual cortex, the visualWulst, (Uchiyama et al. 1987), as well as from the retina(Fig. 1). In contrast to primates, including humans, who

H. Ohno (&) · H. UchiyamaDepartment of Computer and Information Science, Faculty of Engineering, Kagoshima University, Korimoto 1-21-40, Kagoshima 890-0065, Japane-mail: ohno@ics.kagoshima-u.ac.jp

123

340 Exp Brain Res (2009) 194:339–346

move their eyes rapidly to scan the visual scene, birds suchas pigeons, chicks, and quail have long, Xexible necks, andorient their head to the target before pecking (Goodale1983). The present study has, for the Wrst time, recorded theactivity of IO neurons in awake birds with unrestrainedheads using an implanted electrode assembly while record-ing unrestrained head movements using a high-speed videocamera. This enabled us to understand the activity of the IOneurons during normal behavior. This is a key experimentfor understanding the functional signiWcance of this localand transient enhancer of retinal output.

Materials and methods

Experiments were performed on 18 quails (Coturnix japon-ica) obtained from a local commercial supplier. Animalswere treated in accordance with the guidelines for animalusage of the Society for Neuroscience, and all experimentalprotocols were approved by the local committee for AnimalWelfare of Kagoshima University. Animals were anesthe-tized by intramuscular injection of ketamine hydrochloride(3 mg/100 g body weight) and xylazine hydrochloride(0.92 mg/100 g body weight) and placed in a stereotaxicapparatus. All pressure points were anesthetized using lido-caine jelly. The skin over the skull was incised under localanesthesia, and a portion of the skull was opened using adental drill. An electrode assembly was inserted stereotaxi-cally into the ION from the dorsal aspect through the cere-bellum. A hand-made microdrive attached to the electrodeassembly was cemented to the skull (Izawa et al. 2005).Animals recovered overnight and were fed normally on the

following day. The electrode assembly comprised two Wne,H-ML-coated tungsten wires (� = 15 �m; California FineWire, Grover Beach, CA, USA) twisted and glued together,a stainless steel injection needle (30 ga), and a wire connec-tor. The electrode assembly was advanced vertically by themicrodrive. Spikes from IO neurons were ampliWed (headampliWer made with an operational ampliWer (TL084CP),and diVerential ampliWer; Nihon Kohden, Tokyo, Japan;band-pass Wlter, 150–3,000 Hz), and fed to a PC for analy-sis via an A/D converter (1401plus; CED, Cambridge, UK).Spike data were sampled at 83,000 samples/s using Spike2software (CED), and analyzed oV-line using Spike2 andMATLAB (Mathworks, Natick, MA, USA). After recoveryfrom the operation, each animal’s head was left unre-strained during recording sessions, while most of the bodywas covered with a soft felt jacket. Head movements wererecorded using a high-speed video camera (100 fps; VFC-300, Houei, Tokyo, Japan) and a digital video camera(30 fps; DCR-HC40, Sony, Tokyo, Japan), and data werestored in a computer. High-speed video movies and neuro-nal activity recordings were synchronized with DC pulsesand corresponding light pulses emitted from a light-emit-ting diode (LED) lit by the DC pulses. Onset and oVset ofall head movements were logged from the recorded videoimages, and the movements were classiWed into eight direc-tions based on the side in which the recording electrodeswere inserted: upward, downward, ipsilateral, contralateral,ipsilateral upward, contralateral upward, ipsilateral down-ward, and contralateral downward. Receptive Welds of theIO neurons were mapped by moving a hand-held LEDaround the head in the dark, and classiWed into one of thenine sectors of the visual Weld for each eye: upper-anterior,

Fig. 1 The avian retinopetal system: Wber connections (a) and sche-matic representation of the system architecture (b). The retinopetalsystem comprises three serially connected neurons: tecto-IO neurons,isthmo-optic (IO) neurons, and IO target cells. The system receives

both top-down and bottom-up excitatory signals from the avian visualcortex homolog (the visual Wulst) and the retina, respectively. Plussymbol indicate excitatory signals, and asterisk indicates a facilitatorysignal. Dashed lines indicate indirect pathways

+

++

+

* tecto-IO neuron

[retina] retinalganglioncell

IO target cell

[tectum]

[avian "visual cortex"]

[ION]

IO neuron

"attentionfield"

retinaloutput

IOtargetcell

avian "visual cortex"retinal output

IOneuron

tecto-IOneuron

a b

retinopetalmodule

123

Exp Brain Res (2009) 194:339–346 341

upper-central, upper-posterior, middle-anterior, middle-central, middle-posterior, lower-anterior, lower-central, andlower-posterior. The middle-central sector was deWned asthe 30° square sector in the midst of the visual Weld, andother eight outside sectors were deWned accordingly. Thedirections of motions toward the receptive Welds forrecorded IO neurons were deWned as follows: contralateralmotions for receptive Welds in the middle-central sector,contralateral downward motions for receptive Welds in theleft lower-central sector, and downward motions for recep-tive Welds in the lower-anterior sectors. After completion ofthe recording session, the position of the recording elec-trode tip was conWrmed histologically, by passing a current(0.3 mA, 3–5 s), Wxing by intravascular perfusion with tentimes diluted solution of formalin (saturated aqueous solu-tion of formaldehyde), cutting frontal 50-�m sections witha Microslicer (Dosaka, Kyoto, Japan), and staining withhematoxylin. The ION is small (approximately 0.7 £ 0.4 £0.6 mm) in Japanese quail and is located more than 5 mmdeep in the cerebellar surface. Thus, precise targeting of theelectrode tips into the nucleus was diYcult. Histologicalexamination conWrmed that recording sites were within theION proper in four of 18 birds.

Results

We recorded 16 IO neurons in four birds (ten from one birdand two each from three birds); the recording site was veri-Wed histologically as the ION (Fig. 2). Twelve units wererecorded from the right ION in two birds, and 4 U wererecorded from the left ION in two birds. These IO cellsresponded vigorously to moving light stimuli. The recep-tive Welds were mapped using these stimuli, and weredistributed as follows: one in the middle-central sector ofthe visual Weld, nine in the lower-central sector, and six inthe lower-anterior sector. The responses to light stimuli andthe dimensions of the receptive Welds were not diVerentfrom those of the IO neurons in anesthetized and decere-brated animals, which were reported previously (Holdenand Powell 1972; Miles 1972a; Uchiyama and Barlow1994). In contrast to anesthetized animals, IO neurons inawake animals were more active and showed more sponta-neous (self-initiated, rather than stimulus-driven) activity.That is, in awake animals, the IO neurons often Wred in theabsence of visual stimuli or eye–head movements. Sponta-neous activity was observed not only in the middle of headWxation (Fig. 3a) but also before head saccades (Fig. 3b).Figure 3b shows the phasic activity of an IO cell just beforea downward head saccade. In this case, the IO neuronstarted to Wre while the head remained perfectly stationary,about 220 ms before the head saccade. Thus, this presacc-adic activity of IO neurons was not caused by passive

movement of visual images on the retina. Although headsaccades in any direction were not always preceded by suchincreases in IO activity, in most cases the presaccadic activ-ity in one direction of the subsequent saccades was greaterthan that in other directions. For example, in the cellsshown in Figs. 3 and 4, IO activity was observed more fre-quently before head movements in the downward directionthan before movements in other directions. Because thereceptive Welds of these cells were located in the lower-anterior and lower-central sectors of the visual Weld, thesedownward movements turned the head toward the receptiveWelds of cells. This directional speciWcity of activation wasobserved in all 16 cells.

To facilitate the analysis of these complex data, direc-tions of head movements were coarsely classiWed into eightdirections, the amplitudes of movements were disregarded,and the directions toward the receptive Welds of IO neuronswere only roughly deWned. Nevertheless, the association ofpreferential activation of IO neurons during a period ofabout 200 ms before onset of head saccades with movementtoward the receptive Weld was signiWcant (P < 0.01,Student’s t test) (Fig. 5). More precise determination of thedimensions of head saccades might show a greater eleva-tion of activity of the IO neurons before the onset of headsaccades directed at the receptive Welds. Activity of all IOcells was suppressed during the head movements, but onlyweakly, and sometimes IO neurons Wred even during theearly phases of head saccades (Fig. 4a, b). After the headsaccades, activities of IO neurons seem to be elevated tran-siently. This elevation might rebound after saccadic sup-pression or activation by input newly reached from theretina.

Fig. 2 Histological conWrmation of the recording site (7050). Electro-coagulation was performed by passing a current (1 mA, 10 s) throughthe recording electrode after recording sessions. Hematoxylin stain.50-�m frontal section. IV trochlear nucleus, cb cerebellum, tec tectum.Bar 1 mm

123

342 Exp Brain Res (2009) 194:339–346

We analyzed posteriori directional tuning of presaccadicactivity of IO neurons during the 200 ms before the onset ofhead saccades (Fig. 6). Figure 6a shows a polar plot of thesame unit (7050-U10) as in Fig. 4. The unit whose recep-tive Weld was located in the lower-anterior sector showedthe strongest presaccadic activity before contralateraldownward motions. Motion directions that are mostlycorrelated to high priori activity before the onset of head

saccades of IO neurons were obtained by composing activ-ity vectors in eight directions. The directions were close tothe contralateral direction for the unit whose receptive Weldwas in the middle-central sector (Fig. 6b), and were distrib-uted around the contralateral downward direction for unitswhose receptive Welds were in the lower-central sector(Fig. 6c) and around the downward direction for unitswhose receptive Welds were in the lower-anterior sector

Fig. 3 Activity of an isthmo-optic (IO) neuron (7050-U13) duringhead Wxation (a), and before and during a head saccade toward itsreceptive Weld (b) and in another direction (c); the receptive Weld wasin the lower-central sector of the visual Weld. In each part of the Wgure,selected frames from high-speed video images are shown in the upperhalf, and spike activity is shown in the lower half. Records from a–b tob–c are not continuous. Numbers above the images indicate the time at

which the images were captured relative to the time of saccade onset.In a, times represent milliseconds after the onset of the previous headsaccade, and in b and c, times indicate milliseconds before (–) and after(+) the onset of the head saccade (at time = 0). Vertical dashed linesindicate the time in the activity recording at which each image was cap-tured. Horizontal bars in b and c indicate the duration of head saccades

a

b

c

200 300 400 500 600 700

-300 -200 -100 0 100 160 260

-300 -200 -100 0 100 170 270

123

Exp Brain Res (2009) 194:339–346 343

(Fig. 6d). Thus, the presaccadic activity strength of the IOneurons was highly speciWc to the direction of the posteriorisaccadic motions.

Additionally, we recorded the IO neuronal activity indarkness (illuminance <1 lx) and found that IO neuronsshowed intermittent activity during head Wxation withoutany visual stimulus (Fig. 7), supporting the idea that IOneurons can Wre spontaneously (not passively).

Discussion

This is the Wrst report to describe the activity of retinopetalneurons in awake and normally behaving vertebrates, withthe possible exception of a study that used head-restrainedchickens (Marin et al. 1990). As described above, the avianretinopetal system can be regarded as a mechanism forlocal, transient gain enhancement of retinal output. Thus,results of the present study indicate possible presaccadicenhancement of local retinal output. However, presaccadicactivity may be neither eVerent copy nor corollary dis-charge originating in the motor command areas, judgingfrom the loose coupling between the activation of IO

Fig. 4 Activity of an isthmo-optic (IO) neuron (7050-U10) before andduring head saccades in the direction of the receptive Weld, which waslocated in the lower-anterior sector of the visual Weld (a, b) or in otherdirections (c, d). Spike activity is shown in raster representations (a, c)

and time histograms in 20-ms bins (b, d). Times indicate millisecondsbefore (–) and after (+) the onset of the head saccade (vertical line attime = 0). The numbers of head saccades in the direction of the recep-tive Weld and in other directions were 61 and 77, respectively

0

10

20

30

40

50

0

10

20

30

40

50

-300 -200 -100 0 100 200 -300 -200 -100 0 100 200

-300 -200 -100 0 100 200 -300 -200 -100 0 100 200

time to head saccade onset (ms)

aver

age

activ

ity (

spik

es/s

)

aver

age

activ

ity (

spik

es/s

)

head

sac

cade

eve

nts

head

sac

cade

eve

nts

time to head saccade onset (ms)

time to head saccade onset (ms) time to head saccade onset (ms)

a

b

c

d

Fig. 5 Average activity of isthmo-optic (IO) neurons (n = 16) beforeand during head saccades toward their receptive Weld (solid line) andin other directions (dashed line). Bins are 10 ms. Horizontal bar indi-cates the period (¡200 to 0 ms) during which the two values showed astatistically signiWcant diVerence (P < 0.01, Student’s t test). Timesindicate milliseconds before (–) and after (+) the onset of the head sac-cade (vertical line at time = 0). Asterisk next to y-axis indicates theoverall average activity of all IO neurons during all recording sessions(21.8 spikes/s)

-200 -100 0 1000

10

20

30

40

P < 0.01

*

aver

age

activ

ity (

spik

es/s

)

time to head saccade onset (ms)

123

344 Exp Brain Res (2009) 194:339–346

Fig. 6 a Polar plot of average activity of an isthmo-optic (IO) neuron(7050-U10) during 200 ms before the onset of head saccades in eightdirections. Dashed arrow indicates the direction obtained by compos-ing the eight activity vectors. b–d Posteriori motion directions that aremostly correlated to high priori activity of each IO neuron, obtained bycomposing eight activity vectors. Data were classiWed into three

groups, depending on sectors of their receptive Welds: middle-central(b), lower-central (c), and lower-anterior (d). Asterisks in c and d indi-cate the average of each direction. u upward, d downward, c contralat-eral, i ipsilateral, cu contralateral upward, cd contralateral downward,iu ipsilateral upward, id ipsilateral downward

0.5

1 u

cu

c

cd

d

id

i

iu

∗∗

u

cu

c

cd

d

id

i

iu

u

cu

c

cd

d

id

i

iu

u

cu

c

cd

d

id

i

iu

a b

c d

Fig. 7 Activity of an isthmo-optic (IO) neuron (7110-U1) during headWxation in the dark without any visual stimulus. Upper recordingshows spike trains emitted from the IO neuron. Lower images showframes from the digital video image, which was simultaneously record-ed with the unit activity, using infrared “night-shot” mode. Numbers

above the images indicate the time during head Wxation. Verticaldashed lines indicate the time in the activity recording at which eachframe was captured. The illuminance in the light-tight chamber wasless than 1 lx (measured by Konica Minolta CL-200, Tokyo, Japan)

165 330 495 6600

123

Exp Brain Res (2009) 194:339–346 345

neurons and initiation of head saccades. Furthermore,presaccadic activity may not be simply a signal that antici-pates head saccades, as IO neurons Wre in the middle ofeye–head Wxation, as well as before the onset of head sac-cades. As described above, most birds have long, Xexiblenecks, and shift their gaze largely by head movements.Therefore, we monitored head movements in the presentstudy. Although we observed coherent eye movements pre-ceding head movements, we never encountered major,independent eye movements during head Wxation. How-ever, because of technical limitation of the video monitor-ing, we could have failed to detect eye movements of minoramplitude during head Wxation. Using more sensitive mea-surements of eye movements, it must be studied whether IOneurons may modulate their activity around independenteye movements during head Wxation.

Selective visual attention is a mechanism that enablesanimals to rapidly direct their gaze toward objects ofinterest in the visual environment (Itti 2003). Attentioncan shift covertly without eye movements, and covertshifts of attention precede overt eye movements (Livers-edge and Findlay 2000). Theoretical studies of the imple-mentation of visual attention predict the presence of asaliency map, a topographical map represented some-where in the brain that evaluates the salient features of theobjects in the visual scene (Itti and Koch 2001). Thesaliency map is supposed to couple with a global competi-tion mechanism, also known as the winner-take-all(WTA) network, for further topographical tuning of theoutput from the saliency map. Considering these proper-ties, together with the attention models described above,the two-dimensional distribution of the retinopetal mod-ules could represent a saliency map that receives and inte-grates top–down and bottom–up signals. Furthermore,retinopetal modules globally compete among themselves.A few active retinopetal modules may attentionally “illumi-nate” the local retinal areas, or “attention Welds,” that theyinnervate (Fig. 1b). Thus these modules might function asthe “attention spotlight” as proposed by Treisman (1986).

Based on our results, we can hypothesize that the retin-opetal system (i.e., activity of IO neurons) signals covertspatial attention. That is, an active IO neuron projects to aspeciWc local retinal region or “attention Weld,” which isbrought to attention or “illuminated” by that IO activity.When IO activity shifts spontaneously to a diVerent IO neu-ron population, in the middle of head Wxation, it results inactivation of diVerent attention Welds. This could beregarded as covert attentional scanning, or scanning with-out eye–head movements. After several consecutive covertattentional shifts, that is, after migration of activity fromone IO neuron population to another, animals may be ori-ented toward the region that was most recently covertlyattended to. The observation that IO neurons Wre before the

onset of head saccades toward the receptive Welds (attentionWelds) is consistent with this idea.

Because it relies solely upon neural activity, rather thanmechanical head or eye movements, covert attention scan-ning is a rapid and eVective mechanism for visual searchand detection. It would be particularly useful in animalssuch as chickens and quails, whose retinas have wide areasof moderately increased cell density but no distinctivefovea (Ikushima et al. 1986). Rogers and Miles (1972)reported that vision in chicks using the eye contralateral toan ION lesion is impaired, and is severely defective invisual search and object detection. These deWcits in visualsearch and object detection may be caused by loss of thecovert attention scanning process that is normally providedby the retinopetal system. The avian retinopetal system canbe regarded as the tectofugal pathway to the retina becauseof its origin in the tectum.

Recently, the mammalian superior colliculus (SC) wasshown to play a causal role in covert attention shifting and/or target selection for saccades, in addition to being causalin the execution of saccades (Carello and Krauzlis 2004;Cavanaugh et al. 2006; Cavanaugh and Wurtz 2004;Ignashchenkova et al. 2004; McPeek and Keller 2004;Muller et al. 2005). Both the mammalian SC and its avianhomolog, the optic tectum, are likely to play importantroles in attentional selection processes. The avian retinop-etal system may thus provide further evidence for the pre-motor theory of attention (Rizzolatti et al. 1994), whichstates that the same brain mechanisms that underlie the gen-eration of saccades to one part of the visual Weld also con-tribute to the facilitation of visual processing in that part ofthe visual Weld. Physiological studies suggested multiplecandidates for representation of the saliency map in prima-tes, such as the frontal eye Weld (Thompson and Bichot2005), posterior parietal cortex (Kusunoki et al. 2000), pul-vinar nuclei (Robinson and Petersen 1992), and SC (Kustovand Robinson 1996). However, the underlying neuralmechanisms of the saliency map and the WTA networkhave not yet been clariWed in any of these sites. The speci-Wcity and simplicity of cellular organization of the avianretinopetal system (Fig. 1) may make this an ideal modelfor deciphering the cellular basis of subcortical attentionaland/or selection processes. However, mammalian retinop-etal systems are much more diVuse and have diVerent reti-nal targets, and therefore their functional properties may bediVerent from those of birds (Gastinger et al. 2006). Fur-thermore, it remains to be understood why attentional mod-ulation in birds would take place in the retina, rather thanmore centrally as appears to be the case in mammals. Onepossible reason may be the greater structural and functionalcomplexity of the retina, coupled with the lesser complex-ity of the brain, in non-mammalian compared withmammalian species. This could result in heavier reliance

123

346 Exp Brain Res (2009) 194:339–346

upon peripheral processing of visual information in non-mammalian species, as suggested by some authors (Dubin1970).

In conclusion, we found that the IO neurons showed alarge phasic elevation in their activity before head saccadestoward their receptive Weld. These results imply that localretinal output may be enhanced by centrifugal signalsbefore gaze shifts, topographically and in a manner speciWcto the directions of posteriori motions.

Acknowledgments We wish to thank Drs. William K. Stell,Frederick A. Miles, Robert B. Barlow and Masakazu Konishi forvaluable comments on the manuscript. Dr. Stell also kindly edited themanuscript for English usage. This work was partly supported byKAKENHI (#20300095) and funding from Kagoshima University.

References

Carello CD, Krauzlis RJ (2004) Manipulating intent: Evidence for acausal role of the superior colliculus in target selection. Neuron43:575–583

Cavanaugh J, Wurtz RH (2004) Subcortical modulation of attentioncounters change blindness. J Neurosci 24:11236–11243

Cavanaugh J, Alvarez BD, Wurtz RH (2006) Enhanced performancewith brain stimulation: Attentional shift or visual cue? J Neurosci26:11347–11358

Cowan W (1970) Centrifugal Wbres to avian retina. Br Med Bull26:112–118

Dowling JE, Cowan WM (1966) An electron microscope study ofnormal and degenerating centrifugal Wber terminals in the pigeonretina. Z Zellforsch Mikrosk Anat 71:14–28

Dubin MW (1970) The inner plexiform layer of the vertebrate retina:a quantitative and comparative electron microscopic analysis.J Comp Neurol 140:479–505

Gastinger MJ, Tian N, Horvath T, Marshak DW (2006) Retinopetalaxons in mammals: Emphasis on histamine and serotonin. CurrEye Res 31:655–667

Goodale MA (1983) Visually guided pecking in the pigeon (Columbalivia). Brain Behav Evol 22:22–41

Holden AL, Powell TP (1972) The functional organization of the isth-mo-optic nucleus in the pigeon. J Physiol 223:419–447

Ignashchenkova A, Dicke PW, Haarmeier T, Thier P (2004) Neuron-speciWc contribution of the superior colliculus to overt and covertshifts of attention. Nat Neurosci 7:56–64

Ikushima M, Watanabe M, Ito H (1986) Distribution and morphologyof retinal ganglion cells in the Japanese quail. Brain Res 376:320–334

Itti L (2003) Visual Attention. In: Arbib M (ed) The Handbook ofBrain Theory and Neural Networks. MIT Press, pp 1196–1201

Itti L, Koch C (2001) Computational modelling of visual attention.Nature Reviews Neuroscience 2:194–203

Izawa EI, Aoki N, Matsushima T (2005) Neural correlates of the prox-imity and quantity of anticipated food rewards in the ventral stri-atum of domestic chicks. Eur J NeuroSci 22:1502–1512

Kustov AA, Robinson DL (1996) Shared neural control of attentionalshifts and eye movements. Nature 384:74–77

Kusunoki M, Gottlieb J, Goldberg ME (2000) The lateral intraparietalarea as a salience map: the representation of abrupt onset, stimu-lus motion, and task relevance. Vis Res 40:1459–1468

Liversedge SP, Findlay JM (2000) Saccadic eye movements and cog-nition. Trends in Cognitive Sciences 4:6–14

Marin G, Letelier JC, Wallman J (1990) Saccade-related responses ofcentrifugal neurons projecting to the chicken retina. Exp BrainRes 82:263–270

McGill JI, Powell TP, Cowan WM (1966) The retinal representationupon the optic tectum and isthmo-optic nucleus in the pigeon.J Anat 100:5–33

McPeek RM, Keller EL (2004) DeWcits in saccade target selection afterinactivation of superior colliculus. Nat Neurosci 7:757–763

Miles FA (1972a) Centrifugal control of the avian retina. II. ReceptiveWeld properties of cells in the isthmo-optic nucleus. Brain Res48:93–113

Miles FA (1972b) Centrifugal control of the avian retina. III. EVects ofelectrical stimulation of the isthmo-optic tract on the receptiveWeld properties of retinal ganglion cells. Brain Res 48:115–129

Muller JR, Philiastides MG, Newsome WT (2005) Microstimulation ofthe superior colliculus focuses attention witnout moving the eyes.Proc Natl Acad Sci USA 102:524–529

Ramón y Cajal S (1889) Sur la morphologie et les connexions des ele-ments de la retine des oiseaux. Anat Anz 4:111–121

Rizzolatti G, Riggio L, Sheliga BM (1994) Space and Selective Atten-tion. In: Attention and Performance Xv, vol 15, pp 231–265

Robinson DL, Petersen SE (1992) The Pulvinar and Visual Salience.Trends Neurosci 15:127–132

Rogers LJ, Miles FA (1972) Centrifugal control of the avian retina. V.EVects of lesions of the isthmo-optic nucleus on visual behaviour.Brain Res 48:147–156

Thompson KG, Bichot NP (2005) A visual salience map in the primatefrontal eye Weld. Prog Brain Res 147:251–262

Treisman A (1986) Features and objects in visual processing. Sci Am255:114–125

Uchiyama H (1989) Centrifugal pathways to the retina: inXuence of theoptic tectum. Vis Neurosci 3:183–206

Uchiyama H (1999) The isthmo-optic nucleus: A possible neural sub-strate for visual competition. Neurocomputing 26–7:565–571

Uchiyama H, Barlow RB (1994) Centrifugal inputs enhance responsesof retinal ganglion cells in the Japanese quail without changingtheir spatial coding properties. Vision Res 34:2189–2194

Uchiyama H, Ito H (1993) Target cells for the isthmo-optic Wbers in theretina of the Japanese quail. Neurosci Lett 154:35–38

Uchiyama H, Stell WK (2005) Association amacrine cells of Ramon yCajal: Rediscovery and reinterpretation. Vis Neurosci 22:881–891

Uchiyama H, Watanabe M (1985) Tectal neurons projecting to the isth-mo-optic nucleus in the Japanese quail. Neurosci Lett 58:381–385

Uchiyama H, Matsutani S, Watanabe M (1987) Activation of the isth-mo-optic neurons by the visual Wulst stimulation. Brain Res406:322–325

Uchiyama H, Ito H, Tauchi M (1995) Retinal neurones speciWc forcentrifugal modulation of vision. NeuroReport 6:889–892

Uchiyama H, Yamamoto N, Ito H (1996) Tectal neurons that partici-pate in centrifugal control of the quail retina: a morphologicalstudy by means of retrograde labeling with biocytin. Vis Neurosci13:1119–1127

Uchiyama H, Nakamura S, Imazono T (1998) Long-range competitionamong the neurons projecting centrifugally to the quail retina. VisNeurosci 15:417–423

Uchiyama H, Aoki K, Yonezawa S, Arimura F, Ohno H (2004) Retinaltarget cells of the centrifugal projection from the isthmo-opticnucleus. J Comp Neurol 476:146–153

Wallenberg A (1898) Das mediale Opticusbündel der Taube. NeurolZbl 17:532–537

123