Development of Orientation Preference in theMammalian Visual Cortex

Barbara Chapman,1 Imke Godecke,1,2* Tobias Bonhoeffer2

1 Center for Neuroscience, 1544 Newton Court, University of California, Davis, California 95616

2 Max Planck Institut fur Neurobiologie, Am Klopferspitz 18A, 82152 Munchen-Martinsried, Germany

Received 10 May 1999; accepted 20 May 1999

ABSTRACT: Recent experiments have studied thedevelopment of orientation selectivity in normal ani-mals, visually deprived animals, and animals where pat-terns of neuronal activity have been altered. Resultsof these experiments indicate that orientation tuningappears very early in development, and that normalpatterns of activity are necessary for its normal devel-opment. Visual experience is not needed for early devel-

opment of orientation, but is crucial for maintainingorientation selectivity. Neuronal activity and vision thusseem to play similar roles in the development of orien-tation selectivity as they do in the development of eye-specific segregation in the visual system.© 1999 John Wiley

& Sons, Inc. J Neurobiol 41: 18–24, 1999

Keywords:visual cortex; orientation selectivity; oculardominance; activity maps; ferret; development

One of the most conspicuous functional properties ofneurons in the adult visual cortex is their orientationselectivity. How orientation selectivity develops dur-ing early postnatal life, and what role visual experi-ence plays in this process are questions of consider-able interest. Experience-dependent development ofanother response property of cortical cells, oculardominance, has been widely studied, but it is onlyrecently that a substantial amount of new informationabout the development of orientation preference hasbeen gathered. Electrophysiological and optical imag-ing studies have elucidated the timing of the devel-opment of orientation selectivity at the single-celllevel and in orientation preference maps. Pharmaco-logical manipulations and visual deprivation studieshave examined the roles of spontaneous and visuallydriven neuronal activity. These studies show that ori-entation selectivity and its orderly representation inorientation preference maps appear early in develop-

ment. Furthermore, they show that normal neuronalactivity patterns are vital for the development ofproper orientation selectivity. In contrast, visuallydriven activity is not needed for the initial develop-ment of orientation selectivity or orientation maps.Vision does, however, become important for normalmaturation and maintenance of cortical orientationselectivity.

EXPERIMENTAL RESULTS

Early Experiments

Single-Cell Approaches.Efforts to determine whencells in cat visual cortex first show signs of orientationselectivity have proved difficult, yielding somewhatconflicting results. The reported percentages of orien-tation-selective cells in 1-week-old kittens has variedfrom zero (Barlow and Pettigrew, 1971; Pettigrew,1974), through 20–30% (Blakemore and VanSluyters, 1975; Buisseret and Imbert, 1976; Fregnacand Imbert, 1978; Albus and Wolf, 1984), to 100%(Hubel and Wiesel, 1963). Some of this discrepancymay be due to differences in the definitions of “ori-

* Present address:Optical Imaging Europe, Am Klopferspitz19, 82152 Mu¨nchen-Martinsreid, Germany

Correspondence to:B. ChapmanContract grant sponsor: NIH; contract grant number: EY11369Contract grant sponsor: Max-Planck Gesellschafr

© 1999 John Wiley & Sons, Inc. CCC 0022-3034/99/010018-07

18

entation-selective” responses used by the differentauthors. Moreover, cortical cells in young animals arepoorly responsive and habituate rapidly (Hubel andWiesel, 1963), making a precise assessment of neu-ronal response characteristics rather difficult.

Functional Organization. In their initial studies onthe tangential organization of the cat visual cortexHubel and Wiesel (1962) described the orderly ar-rangement of cells into orientation preference maps.Studying how and when these maps develop in theyoung animal has proven to be an even greater tech-nical challenge than studying the development of ori-entation selectivity at the single-cell level. Single-unitelectrophysiological mapping experiments providevery limited information about orientation preferencemaps even in adult animals, and are impractical invery young animals. 2-Deoxyglucose labeling tech-niques have provided some information about earlyorientation maps (LeVay et al., 1978, 1980; Des Ros-ier et al., 1978; Thompson et al., 1983), but the factthat such studies cannot be performed chronically,together with the individual differences in the layoutsof the maps between animals, makes these resultsdifficult to interpret.

Newer Experimental Approaches

Recent advances, including the use of the ferret as anexperimental animal and the use of the optical imag-ing technique, have helped to mitigate the technicaldifficulties of studying the development of orientationpreference in visual cortex and have led to new in-sights in this field.

Ferrets have a visual system which is quite similarto that of the cat (Law et al., 1988), but they are borndevelopmentally approximately 3 weeks younger(Linden et al., 1981). Therefore, ferrets provide arelatively hardy electrophysiological preparation dur-ing the time when orientation-specific responses aredeveloping, and are thus an ideal animal to study thedevelopment of orientation preference in young ani-mals.

Optical imaging of intrinsic signals (Grinvald etal., 1986; Bonhoeffer and Grinvald, 1996) is a tech-nique used to study the functional tangential organi-zation of the cerebral cortex. Because of its relativenoninvasiveness [demonstrated early on by Frostig etal. (1990) by imaging through the closed skull], it isideally suited for studying cortical maps in younganimals. Most important, chronic optical imaging forthe first time has allowed the development of orien-tation preference maps to be followed in individualanimals.

In addition to those major improvements, a number

of other new techniques have become available topharmacologically and electrically influence activityof cells in the developing visual system. The combi-nation of these methodological advances has over thepast 5–10 years allowed more precise experimentalassessment of the factors controlling the emergence oforientation preference in the visual cortex.

Normal Development of Orientation SelectivityStudied at the Single-Cell Level.Early postnatal de-velopment of orientation selectivity was studied innormal pigmented ferrets (Chapman and Stryker,1993). The earliest age at which visually driven ac-tivity was observed in ferret visual cortex was post-natal day (P)23, comparable to P2 in the cat. Thisdevelopmental stage is considerably earlier than thetime when it was possible to record neuronal activityin kitten cortex. Nevertheless, at this age good orien-tation selectivity was found in some ferret corticalcells. Most neurons at this age, however, showedunoriented responses. The percentage of oriented cellsremained constant until the end of postnatal week 5(corresponding to week 2 in kitten), when more ori-ented cells could be observed. The number of orientedcells continued to increase until postnatal week 7, atwhich time the proportion of orientation-selectiveneurons was indistinguishable from that seen in theadult.

Normal Development of Orientation Maps.The tim-ing of the development of orientation preference mapsdoes not necessarily follow that of the appearance ofthe first oriented responses in visual cortex. For anorientation map to be seen with optical imaging, asubstantial number of oriented cells is required. Inaddition, it is conceivable that neurons first acquiretheir orientation selectivity in a spatially unorganizedfashion, and that the orderly representation of theorientation preference map is only achieved in a sec-ond developmental step.

The emergence of orientation maps in the visualcortex was first studied in ferrets (Chapman et al.,1996). Figure 1 shows an example of a chronic opticalimaging experiment following the development oforientation maps in an individual animal. In this rep-resentative example (for others, see Chapman et al.,1996), it is apparent that while the maps get strongerover time, their general structure (i.e., size, shape,location of the iso-orientation domains) changes verylittle during the course of development.

Later studies in kittens (Go¨decke et al., 1997)yielded similar results. In both ferret and cat, thetiming of the map development lagged behind thedevelopment of orientation selectivity at the singlecell level (Fig. 2).

Orientation Preference in Mammalian Visual Cortex 19

Effects of Altered Patterns of Neuronal Activity.Inaddition to studies of normal development, manipu-lations designed to disrupt spontaneous or visuallydriven neuronal activity have proved useful for un-derstanding how orientation selectivity develops inthe visual cortex.

The critical dependence of orientation selectivitydevelopment on neuronal activity has been demon-strated by experiments in which neuronal activity inthe cortex was completely abolished. This wasachieved by intracortical infusion of the sodium chan-nel blocker tetrodotoxin (TTX) during the time whencortical orientation selectivity normally develops.This activity blockade prevented the development ofcortical orientation selectivity (Chapman and Stryker,1993). Ferrets treated with TTX during the period oforientation maturation (from postnatal week 4 throughpostnatal week 7) maintained the immature distribu-

tion of orientation selectivities typical of postnatalweek 4 animals.

A more subtle pharmacological manipulation ofactivity consists of disrupting the normal equilibriumbetween on-center and off-center responses in thevisual system. This can be achieved by silencingon-center retinal ganglion cells during developmentwith daily intravitreal injections of the glutamate an-alog 2-amino-4-phosphonobutyrate (APB). This pro-cedure, like cortical application of TTX, “freezes”orientation selectivity in an immature state (Go¨deckeand Chapman, 1998). Interestingly, when ferrets weretreated with an APB dosage which blocked on-centeractivity only half the time, normally organized(though weak) orientation maps developed (Go¨deckeand Chapman, 1998). This indicates that discontinu-ous periods of “normal” activity are sufficient to pro-duce a normal orientation map.

Figure 1 Development of orientation maps revealed by chronic optical imaging. Single-conditionorientation maps at four different ages in one animal. Each row of the figure shows orientation mapsrecorded in the left primary visual cortex of one ferret at the age [postnatal day (p)] indicated at theleft of the row. Each column of single-condition maps shows orientation maps recorded in responseto a particular orientation of a moving square wave grating (0°5 horizontal). For each map, caudalis up and medial is to the left. The curve in the upper left corner of each map indicates the locationof the caudal pole of the cortex behind which the skull remained intact over the cerebellum. In thisexample, the first clear orientation maps can be seen at postnatal day 38. Individual iso-orientationdomains remain stable over time and do not change their position, shape, or size. Scale bar5 2 mm.

20 Chapman, Go¨decke, and Bonhoeffer

Normal patterns of activity have also been dis-rupted by chronic electrical stimulation of the opticnerve during development (Weliky and Katz, 1997).This overrides intrinsic patterned neuronal activityoriginating from the retina (Meister et al. 1991; Wongand Oakley, 1996) by artificially increasing the cor-relation between inputs to the cortex. The stimulationprotocol, which activated the visual pathway for 10%of the time while leaving activity normal for 90% ofthe time, radically reduced the orientation selectivityof individual cells, while leaving the overall pattern oforientation maps in the cortex intact (Weliky andKatz, 1997).

Effects of Altered Visual Experience.It must havebeen surprising to see that a newborn monkey without

any visual experience shows well-developed orienta-tion selectivity (Hubel and Wiesel, 1974). This studydemonstrated that in the monkey, visual experience isnot needed for the initial development of orientationselectivity. Indeed, later studies in cat (Blakemore andVan Sluyters, 1975; Buisseret and Imbert, 1976) andferret (Chapman and Stryker, 1993) showed that evenduring binocular visual deprivation some degree ofsingle-cell orientation selectivity does develop. Figure3 demonstrates that ferrets binocularly deprived frombefore the time of natural eye opening through the 8thto 13th postnatal week showed significantly betterorientation tuning than immature animals. However,these animals showed significantly poorer orientationtuning than age-matched controls. In this study, noshort-term binocular deprivations were performed, so

Figure 2 Comparison of the development of orientation tuning assessed by optical imaging andelectrophysiology. Orientation tuning assessed electrophysiologically from single-unit recordingscompared with optical imaging of the development of orientation tuning. Optical imaging data fromChapman et al. (1996) (crosses); single-unit data from Chapman and Stryker (1993) (diamonds).The orientation selectivity index for electrophysiological data is calculated from the Fouriertransform of the orientation tuning histogram recorded for each neuron. It equals the amplitude ofthe second harmonic component normalized by dividing by the sum of the DC level and theamplitude of the second harmonic component and multiplying by 100. The orientation tuning for theoptical imaging data is the median length of the vectors in the polar maps. The solid curve indicatesthe best-sigmoid-fit curve through electrophysiological data, while the dashed curve is the best-sigmoid fit from the optical imaging data. The mean of the best-fit sigmoid for the electrophysio-logical data is 4 days earlier (P33.4) than the mean for the optical imaging data (P37.4).

Orientation Preference in Mammalian Visual Cortex 21

it is not clear to what extent these results were due toeffects of deprivation on the initial emergence oforientation and to what extent they were due to effectson the maintenance of orientation selectivity.

Visual experience is also not needed for the initialdevelopment of cortical orientation maps. Early mapsare seen in ferret visual cortex before the time ofnatural eye opening (Chapman et al., 1996), and nor-mal orientation maps develop in kittens binocularlydeprived for the first 3 weeks of life (Go¨decke et al.,1997; Crair et al., 1998). However, longer periods ofbinocular deprivation cause degradation of orientationpreference maps (Crair et al., 1998), indicating thatvisual experience is needed for their maintenance,consistent with prior electrophysiological results(Blakemore and Van Sluyters, 1975; Buisseret andImbert, 1976).

These studies show that the initial development oforientation selectivity does not require visual experi-ence. However, visual experience is needed for main-tenance and maturation of both orientation selectivityat the single-cell level (Chapman and Stryker, 1993)and orientation preference maps (Crair et al., 1998).

It has also become clear that visual experience isnot responsible for the alignment of orientation mapsbetween the two eyes. Kittens raised under a reverse-suture protocol such that the two eyes never had

visual experience at the same time nevertheless de-veloped nearly identical orientation preference mapsfor the two eyes (Go¨decke and Bonhoeffer, 1996).

DISCUSSION

The detailed analysis of the development of orienta-tion selectivity at the single-cell level which has beenpossible in the ferret has raised new questions abouthow cortical orientation develops. The presence of afew orientation selective cells very early in develop-ment, along with the fact that the percentages of tunedcells remain constant for more than a week after thecortex first becomes responsive, suggests that theremay be two different mechanisms underlying orien-tation selectivity in the young cortex. Nothing isknown so far about the development of the earliestoriented cells in the cortex. Further experimentation isneeded to determine whether the development of thisearly orientation selectivity is activity dependent, tostudy the earliest organization of orientation prefer-ence across the cortex, and to discover what neuronalconnections may underlie the early orientation-tunedresponses.

Later development during the fifth postnatal week,when orientation selectivity starts to mature in the

Figure 3 Cumulative histogram showing orientation selectivity index distributions for normalimmature ferrets (postnatal week 4;n 5 4 animals), normal mature ferrets (postnatal week 7 throughadult;n 5 16 animals), and binocularly deprived ferrets (lid sutured from before the time of naturaleye opening through recording sessions on P55, P56, P66, P87, P88, and P90;n 5 6 animals). Thethree distributions are statistically distinct; Mann–WhitneyU test, p , .001). Data are fromChapman and Stryker (1993).

22 Chapman, Go¨decke, and Bonhoeffer

ferret, is known to be dependent on neuronal activity,and seems to proceed normally only when there arenatural patterns of on- and off-center input activity.These results are consistent with computational mod-els of cortical orientation selectivity developmentwhich rely on correlations and anticorrelations in theactivities of on- and off- center inputs to cortex toproduce oriented cortical cell receptive fields (Miller,1992, 1994; Miyashita and Tanaka, 1992).

Because there is little maturation of orientation inferret layer IV cells (Chapman and Stryker, 1993), itis possible that the inputs from the lateral geniculatenucleus afferents are already arranged with receptivefields aligned in space (Chapman et al., 1991; Reidand Alonso, 1995; Ferster et al., 1996) creating ori-entation selectivity in layer IV as soon as the cortex isvisually responsive. The maturation of orientation se-lectivity which occurs later in layer II/III and then inlayer VI could be due first to the refinement of patchyconnections in the supragranular layers (Luhman etal., 1986; Callaway and Katz, 1990; Ruthazer andStryker, 1996) and then to the maturation of connec-tions between supra- and infragranular layers.

Chronic imaging studies show that orientationmaps develop normally in the absence of vision, in thesame way that single-cell studies show that visualexperience is not needed for the initial appearance oforientation preference. The functional layout of theorientation maps, including the typical map structurewith iso-orientation domains forming “pinwheels”(Bonhoeffer and Grinvald, 1991), also develops with-out any visual experience. Proper maintenance of bothindividual cell orientation selectivity and orientationpreference maps, however, only occurs if visual inputis present.

Once development has proceeded to the pointwhere orientation preference maps can be seen usingoptical imaging, the tangential arrangement of orien-tation preferences across cortex is remarkably stable.In both ferrets and kittens, orientation maps remainstable both during normal development (Chapman etal., 1996) and under conditions of altered visual ex-perience (Go¨decke and Bonhoeffer, 1996). Therefore,the layout of orientation preference maps is unaf-fected by the massive rearrangements of geniculocor-tical afferents which occur during development orplasticity of ocular dominance columns. Since genicu-locortical afferents are thought to substantially con-tribute to the determination of orientation selectivity(Chapman et al., 1991; Reid and Alonso, 1995; Fer-ster et al., 1996), this map stability might appearsurprising. However, there is no reason to expect thatthe loss of one eye’s inputs to a given cortical domain(either during normal segregation of afferents intoocular dominance columns or during monocular de-

privation) should affect the arrangement of the othereye’s geniculate inputs to that domain. In addition, thepatchy intracortical connections already establishedduring this time may help further to stabilize theorientation maps.

The development of orientation selectivity in thecortex does not require visual experience, but is crit-ically dependent on normal patterns of spontaneousneuronal activity. In this regard, development of ori-entation selectivity is similar to development of oculardominance in the cortex where normal developmentof columns is seen in binocularly deprived animals(Hubel and Wiesel, 1965; Wiesel and Hubel, 1974),while development is prevented by neuronal activityblockade (Stryker and Harris, 1986) or synchronizedactivity from the two eyes (Stryker and Strickland,1984). As in the case of ocular dominance (Blake-more and Van Sluyters, 1975; Fregnac and Imbert,1978), visual experience is necessary for the mainte-nance of orientation selectivity in the cortex.

From the experiments discussed above, it is appar-ent that proper neuronal activity is needed for theestablishment of orientation and ocular dominancemaps. However, it is unclear whether activity patternscontrol the individual layout of these maps or whetherintrinsic or even genetic factors are the main determi-nants of the functional layout. One attractive andfeasible approach to distinguish between these possi-bilities is to record from genetically identical individ-uals and assess the similarity of their orientation andocular dominance maps.

Financial support was provided by NIH Grant EY11369,to BC, and the Max-Planck Gesellschaft, to TB.

REFERENCES

Albus K, Wolf K. 1984. Early postnatal development ofneuronal function in the kitten’s visual cortex: a laminaranalysis. J Physiol (Lond) 348:153–185.

Barlow HB, Pettigrew JD. 1971. Lack of specificity ofneurones in the visual cortex of young kittens. J Physiol(Lond) 218:98–100.

Blakemore C, Van Sluyters RC. 1975. Innate and environ-mental factors in the development of the kitten’s visualcortex. J Physiol (Lond) 248:663–716.

Bonhoeffer T, Grinvald A. 1996. Optical imaging based onintrinsic signals: the methodology. In: Toga A, MazziottaJC, editors. Brain mapping, the methods. San Diego, CA:Academic Press. p 55–97.

Buisseret P, Imbert M. 1976. Visual cortical cells: theirdevelopmental properties in normal and dark-reared kit-tens. J Physiol (Lond) 255:511–525.

Callaway EM, Katz LC. 1990. Emergence and refinement of

Orientation Preference in Mammalian Visual Cortex 23

clustered horizontal connections in cat striate cortex.J Neurosci 10:1134–1153.

Chapman B, Stryker MP. 1993. Development of orientationselectivity in ferret visual cortex and effects of depriva-tion. J Neurosci 13:5251–5262.

Chapman B, Stryker MP, Bonhoeffer T. 1996. Developmentof orientation preference maps in ferret primary visualcortex. J Neurosci 16:6443–6453.

Chapman B, Zahs KR, Stryker MP. 1991. Relationship ofcortical cell orientation selectivity to alignment of recep-tive fields of the geniculocortical afferents that arborizewithin a single orientation column in ferret visual cortex.J Neurosci 11:1347–1358.

Crair MC, Gillespie DC, Stryker MP. 1998. The role ofvisual experience in the development of columns in catvisual cortex. Science 279:566–570.

Des Rosiers MH, Sakurada O, Jehle J, Shinohara M,Kennedy C, Sokoloff L. 1978. Demonstration of func-tional plasticity in the immature striate cortex of themonkey by means of14C-deoxyglucose method. Science200:447–449.

Ferster D, Chung S, Wheat H. 1996. Orientation selectivityof thalamic input to simple cells of cat visual cortex.Nature 380:249–252.

Fregnac Y, Imbert M. 1978. Early development of visualcortical cells in normal and dark-reared kittens: the rela-tionship between orientation selectivity and ocular dom-inance. J Physiol (Lond) 278:27–44.

Frostig RD, Lieke EE, Ts’o DY, Grinvald A. 1990. Corticalfunctional architecture and local coupling between neu-ronal activity and microstimulation revealed by in vivohigh-resolution optical imaging of intrinsic signals. ProcNatl Acad Sci USA 87:6082–6086.

Godecke I, Bonhoeffer T. 1996. Development of identicalorientation maps for two eyes without common visualexperience. Nature 379:251–254.

Godecke I, Chapman B. 1989. Effects of on-center retinalganglion cell blockade on development of visual corticalresponses. Soc Neurosci Abstr 24:1051.

Godecke I, Kim D-S, Bonhoeffer T. 1997. Development oforientation preference maps in area 18 of kitten visualcortex. Eur J Neurosci 9:1754–1762.

Grinvald A, Lieke EE, Frostig RD, Gilbert CD, Wiesel TN.1986. Functional architecture of cortex revealed by opti-cal imaging of intrinsic signals. Nature 324:361–364.

Hubel DH, Wiesel TN. 1962. Receptive fields, binocularinteraction and functional architecture in the cat’s visualcortex. J Physiol (Lond) 160:106–154.

Hubel DH, Wiesel TN. 1963. Receptive fields of cells instriate cortex of very young, visually inexperienced kit-tens. J Neurophysiol 26:994–1002.

Hubel DH, Wiesel TN. 1965. Binocular interaction in striatecortex of kittens reared with artificial squint. J Neuro-physiol 28:1041–1059.

Law MI, Zahs KR, Stryker MP. 1988. Organization ofprimary visual cortex (area 17) in the ferret. J CompNeurol 278:157–180.

LeVay S, Stryker MP, Shatz CJ. 1978. Ocular dominancecolumns and their development in Layer IV of the cat’svisual cortex: a quantitative study. J Comp Neurol 179:223–244.

LeVay S, Wiesel TN, Hubel DH. 1980. The development ofocular dominance columns in normal and visually de-prived monkeys. J Comp Neurol 191:1–51.

Linden DC, Guillery RW, Cucciaro J. 1981. The dorsallateral geniculate nucleus of the normal ferret and itspostnatal development. J Comp Neurol 203:189–211.

Luhmann HJ, Martinez Millen L, Singer W. 1986. Devel-opment of horizontal intrinsic connections in cat striatecortex. Exp Brain Res 63:443–448.

Meister M, Wong ROL, Baylor DA, Shatz CJ. 1991. Syn-chronous bursts of action potentials in ganglion cells ofthe developing mammalian retina. Science 17:939–943.

Pettigrew JD. 1974. The effect of visual experience on thedevelopment of stimulus specificity by kitten corticalneurones. J Physiol (Lond) 237:49–74.

Miller KD. 1992. Development of orientation columns viacompetition between ON- and OFF-center inputs. Neu-roreport 3:73–76.

Miller KD. 1994. A model for the development of simplecell receptive fields and the ordered arrangement of ori-entation columns through activity-dependent competitionbetween on- and off-center inputs. J Neurosci 14:409–441.

Miyashita M, Tanaka S. 1992. A mathematical model forthe self-organization of orientation columns in visualcortex. Neuroreport 3:69–72.

Reid RC, Alonso JM. 1995. Specificity of monosynapticconnections from thalamus to visual cortex. Nature 378:281–284.

Ruthazer E, Stryker MP. 1996. The role of activity in thedevelopment of long-range horizontal connections in area17 of the ferret. J Neurosci 16:7253–7269.

Stryker MP, Harris WA. 1986. Binocular impulse blockadeprevents formation of ocular dominance columns in thecat’s visual cortex. J Neurosci 6:2117–2133.

Stryker MP, Strickland SL. 1984. Physiological segregationof ocular dominance columns depends on the pattern ofafferent electrical activity. Invest Opthalmol Vis Sci25(Suppl):278.

Thompson ID, Kossut M, Blakemore C. 1983. Developmentof orientation columns in cat striate cortex revealed by2-deoxyglucose autoradiography. Nature 301:712–715.

Weliky M, Katz LC. 1997. Disruption of orientation tuningin visual cortex by artificially correlated neuronal activ-ity. Nature 386:680–685.

Wiesel TN, Hubel DH. 1974. Ordered arrangement of ori-entation columns in monkeys lacking visual experience.J Comp Neurol 158:307–318.

Wong ROL, Oakley DM. 1996. Changing patterns of spon-taneous bursting activity of on and off retinal ganglioncells during development. Neuron 16:1087–1095.

24 Chapman, Go¨decke, and Bonhoeffer