Neurophysiological recordings in freely moving monkeys

Methods 38 (2006) 202–209

www.elsevier.com/locate/ymeth

Neurophysiological recordings in freely moving monkeys

Ning Lei Sun a,d, Yan Lin Lei a, Byoung-Hoon Kim b, Jae-Wook Ryou c, Yuan-Ye Ma a, Fraser A.W. Wilson a,¤

a Laboratory of Primate Neuroscience, Kunming Institute of Zoology, The Chinese Academy of Sciences, 32 Jiaochang Donglu, Kunming, Yunnan 650223, China

b Department of Physiology, University of Wisconsin, USAc Department of Neurology, Cornell University, USA

d Graduate School of the Chinese Academy of Sciences, Beijing 100039, PR china

Accepted 15 September 2005

Abstract

Recordings of neuronal activity in freely moving rats are common in experiments where electrical signals are transmitted using cables.Such techniques are not common in monkeys because their prehensile abilities are thought to preclude such techniques. However, analysisof brain mechanisms underlying spatial navigation and cognition require the subject to walk. We have developed techniques for record-ings in freely moving monkeys in two diVerent situations: a 5£ 5 m testing laboratory and in a 50 m2 open Weld environment. Neuronalsignals are sent to ampliWers and data acquisition systems using cables or telemetry. These techniques provide high quality recordings ofsingle neurons during behaviors such as foraging, walking, and the performance of memory tasks and thus provide a unique opportunityto study primate behavior in a semi-natural situation.© 2006 Elsevier Inc. All rights reserved.

Keywords: Allocentric space; Egocentric space; Hippocampus; Prefrontal cortex primate; Putamen; Rhesus

1. Introduction

The vast majority of neurophysiological studies inbehaving monkeys use techniques originated by EdwardEvarts (e.g. [3]) and remain very similar to those pioneeringstudies. For example, standard methods include immobili-zation of the head and a hydraulically driven drive to movea microelectrode. These techniques are well suited to study-ing many forms of brain function, but certain classes ofbehaviors and cognitive abilities such as learning about thespatial organization of the environment require a diVerentapproach. O’Keefe and Dostrovsky [14] pioneered studiesin recording hippocampal activity in freely moving ratswith their original description of place cells. Many subse-quent studies have conWrmed the original observations:

* Corresponding author.E-mail address: [email protected] (F.A.W. Wilson).

1046-2023/$ - see front matter © 2006 Elsevier Inc. All rights reserved.doi:10.1016/j.ymeth.2005.09.018

place cells are neurons that Wre selectively when rats walkthrough a part of, but not all of a familiar environment.Such cells are candidates for providing information about a‘cognitive map’ of the environment and strongly suggestthat the rat hippocampus is involved in the analysis ofspace [15]. In contrast, the functions of the non-human pri-mate hippocampus are still controversial—it is simply notclear what function the hippocampus contributes. Ourapproach to this controversy has been to undertake studiesin monkeys that replicate the basic paradigm established byO’Keefe. Other investigators have also adapted the freelymoving rat paradigms to monkeys [16,17].

Like rats, non-human primates are foragers and travelconsiderable distances in search of food. Their feedingbehavior is opportunistic [21], and is characterized by sea-sonal variations in the many diVerent foods that they sample[6,7]. One might expect that primate survival is mediated bythe development of brain structures that provide the visualand spatial abilities to enable foraging, and laboratory

N.L. Sun et al. / Methods 38 (2006) 202–209 203

studies amply attest to the ability of monkeys to learn aboutnew food sources [22]. It is striking then that typical labora-tory research with monkeys rarely studies foraging behavioras a means to examine fundamental processes of learningand memory. Part of the explanation for this neglect is thatproper facilities for studying spatial memory are technicallydiYcult to develop and to place under experimental control.However, it is clear that brain mechanisms in humans arespecialized for mapping the environment. For example,Habib and Sirigu [5] describe the loss of familiarity for placesfollowing brain damage. Although progress has been madein using brain imaging techniques to analyse these mecha-nisms in people [12], recordings of neuronal activity couldcontribute to studying the neural bases of learning about thespatial arrangements of cues in the environment by develop-ing techniques for measuring behavior and neuronal activityin freely moving monkeys engaged in walking throughextended spatial environments. In this article we documenttwo diVerent approaches to studying neuronal activity infreely moving monkeys walking in extensive spatial environ-ments. In one paradigm we describe procedures for record-ings using cables and a commutator made in a testing roomthe size of a typical research laboratory. In a second para-digm we describe techniques for telemetric recordings inmonkeys walking within a 50 m2 open Weld environment.

2. Description of methods

2.1. General behavioral and safety issues

Rhesus monkeys (Macaca mulatta) are preferred toother species of monkeys (but see [10]), in part because theyare intelligent and eminently trainable, are very resilient,work well on a training schedule, and because their anat-omy and physiology is well known and in many respectscomparable to people. A drawback is that rhesus monkeyscan be relatively aggressive, and can inXict severe injuries.Accordingly, great care must be taken to avoid injury dur-ing freely moving monkey experiments. We have found thatmonkeys are very cooperative when they have learned thatthey will not be harmed and their testing regimen is routine.Under these conditions, monkeys can be lead around onleashes (Fig. 1) without incident although unexpectedevents can cause monkeys to respond aggressively.

We prefer to begin training in juvenile (2–3kg) monkeys,though we have successfully trained an 8-year-old 12kg rhe-sus that had previously lived in a semi-natural environmentwith conspeciWcs in a monkey corral at the Kunming PrimateResearch Centre (www.kiz.ac.cn/introduction/map2.htm).Once trained, monkeys can participate in experimental workfor up to 8 years, can be surgically implanted for neurophysi-ological studies, and may be retired at the end of the experi-ments in good health and weighing 16kg. At this point; theimplants are surgically removed and a skin graft is used toreplace any defect. Recording locations in the brain can beidentiWed with MRI scans and X-radiographs, thus obviatingthe need for histological examination of the brain.

2.2. A laboratory environment for testing freely moving monkeys

2.2.1. The testing rooms and feedersThe testing room measures approximately 5.5£6 m; the

Xoors and walls are gray-green. Four feeders [24] are placedmidway along each wall; the distance between the lids of thefeeders is approximately 2.1m. Fig. 2A shows the dimen-sions, task contingencies and equipment in the testing room.

In the centre of the room is a post (0.5 m high) with arotating spindle at the top. Mounted on the spindle is ahousing that contains a retractable dog leash (Flexi Classic1–5), which rotates as the monkey walks around the room.This leash is suYciently long (»2 m) to allow the monkeysto comfortably sit and forage at the four feeders but notlong enough to allow them to climb the feeders.

The feeders are 0.5 m high, with a 5£ 5 cm lid (covering a2£4 cm slot housing a piece of food) that can be raised,triggering a computer to deliver a reward [24]. The feederswere designed so that eYcient retrieval required the mon-keys to sit before opening the feeder lid. Monkeys usuallyuse one hand to raise the lid and the other to retrieve thefood. This skill requires good visuo-motor coordination,which took several days to learn. Each feeder is able todeliver 50 rewards (e.g., 1/8th grape, apple morsel, 1/2 pea-nut, raisin) per session. Feeders have several notable fea-tures, a white/blue LED (to identify it as the current target),a loudspeaker (to indicate a behavioral error), and a lid(with a magnetic sensor, RadioShack 49–496) that requiredlifting to initiate food delivery. Actuation of the sensor senta signal to a laboratory computer (Pentium 133, DOS oper-ating system; running TEMPO software, ReXective Com-puting, Inc.) when the lid is opened. Fig. 1C shows one ofthe feeders with a curtain behind it (see also http://w3.ari-zona.edu/~primate/).

The curtains running behind the feeders are used toreduce the asymmetry of the room so that it is eVectively a

Fig. 1. Preliminary training for freely moving monkeys in the garden atthe Kunming Institute of Zoology. Monkeys wear a collar on the neckwhich is attached to a leash drawn through a plastic tube.

204 N.L. Sun et al. / Methods 38 (2006) 202–209

square (4.9£ 4.9 m). Curtains provide a largely homoge-neous visual environment and thus the view of the testingroom is largely the same where ever the monkey happens tobe looking. This measure is designed to reduce the visualcues that inXuence neuronal activity. Fluorescent lightswere concealed in soWts close to the ceiling (4.5 m high)along all four walls. A chamber (1 m wide£ 1 mlong£ 0.7 m deep) is located in the center of the room’s ceil-ing. It contained a color video camera (American DynamicsADC762 1/3�) directed vertically down to record the mon-key’s behavior, and an Air Flyte 72 channel slip-ring com-mutator (PN 1001498/72) that is modiWed (DragonXyResearch and Development, Inc.) by providing a servo-con-trolled torque motor. The commutator is used to carryneurophysiological signals to power ampliWers in an adja-cent control room. Investigators monitored equipment andthe monkey’s behavior from the control room.

2.2.2. Behavioral tasksMonkeys were trained on several visual- and memory

guided tasks in the testing laboratory [8]. One useful para-digm is the freely moving monkey version of the delayedalternation task, the performance of which is dependentupon the dorsolateral prefrontal cortex [20] and fornix [13].In this task, the monkeys alternated between three of thefour feeders in a stereotyped sequence (Fig. 2). There weretwo versions of this task. In the North task, monkeys alter-nated in the following way, west-north-east-north-west, etc.In the South task, the monkey alternated east-south-west-south-east, etc. Thus, there were eight possible pathsbetween the feeders in the two versions of the task. Thesetasks required the monkey to travel between the feeders in astereotyped sequence. Each path is taken approximately10–20 times during an experiment (replication is essentialfor neurophysiological studies) and the monkeys achievedaverage speeds of about 1 m/s, with peak speeds of 3–5 m/sbetween the feeders. In a 2 choice version of the delayedalternation task, the monkey alternated between the north

and south, or between the east and west feeders. Recordingsof the visual- and memory guided tasks were performedsequentially in the same testing room.

2.2.3. Plotting head position and direction in the testing roomOur techniques involve the chronic (2–6 months)

implantation of recording electrodes in the brain. Surgicaland implant construction techniques are standard [3] butuse titanium nuts embedded within the dental acrylic sothat bolts can be used to hold a helmet on the implant. Thehelmets (Fig. 3) were made of thermoplastic splinting sheets(Smith & Nephew EzeForm), and are used to protect elec-trodes [23] and preampliWers, and to allow the placement ofLED arrays used to monitor the location of the monkeyswithin the testing room. The LED arrays were mounted onthe helmet so that the LEDs were displaced 4.5� left (red)and right (blue) of the centre of the helmet. The locations ofthe LEDs were tracked by a computer (Cheetah, Neura-lynx, NL) interfaced with a ceiling-mounted video cameraat a temporal resolution of 60 frames/s with 640£ 480 pixelresolution provided by the NTSC signal. The area withinwhich the monkeys walk is represented as a 100£100 ele-ment grid where 1 pixelD7 cm2 (PrimaTracker, Kim et al.,2004a; laboratory website: http://w3.arizona.edu/~primate/).Neuronal Wring rate is obtained by dividing the number ofspikes emitted by the amount of time spent in a particularlocation (pixel). The software is used to determine the cen-tre of the monkey’s head (location), the direction in whichthe head is facing, and the speed and angular velocity ofmovements of the head.

A second method for analysing head movements duringthe memory tasks is to use cameras that were mounted insidethe helmets [9]. This form of data provides comparable infor-mation about the viewing perspective of the monkeys butwithout adequate software is harder to interpret than recon-struction of head movements using the LED arrays. A movieof the scene from the head camera can be seen at the labora-tory website (http://w3.arizona.edu/~primate/).

Fig. 2. The freely moving monkey testing laboratory. (A) Layout of the laboratory.The dashed lines indicates the shortest pathways between the feeders.The path sequences are shown in part b. C/P refers to the position of the camera and commutator mounted in the ceiling. (B). Actual traces of the mon-key’s movements in the testing room, obtained by plotting location as a function of head position. Typically, monkeys take a curved path between feeders;they rarely walk to the center of the room. (C) The south feeder. All feeders are identical, providing a homogeneous testing environment for neurophysio-logical experiments. The curtains running behind the feeders also contribute to this homogeneity.

N.L. Sun et al. / Methods 38 (2006) 202–209 205

2.2.4. Data acquisition and behavioral control systemThe Cheetah system (Neuralynx Inc.) is used to acquire

neuronal waveforms. This system was developed by CaseyStengel and KristoV Agiello based on the DOS-basedBrainWave system, which was originally designed by Sten-gel. The original Cheetah system was implemented on a SunUltraSPARC 1 computer running the Solaris 7 operatingsystem. This platform was believed to provide the necessaryprocessing power and the original speciWcations envisagedtwo animal tracking devices. However, the rapid develop-ment and increasing power of CPUs and a stable operatingsystem (Windows NT, 2000) made it possible to port theCheetah system to an Intel processor-based system. TheCheetah action potential discrimination concept wasdesigned to analyse waveform data acquired using elec-trode bundles in the tetrode conWguration.

Action potentials are recorded diVerentially, buVeredwith a NL headstage-54 ampliWer (Fig. 3), sent by cable to acommutator and then to power ampliWers (NL 8). Formonkeys, we use 36 gauge insulated wires to make a multi-channel cable. The ampliWcation system is controlled byCheetah software which selects waveforms for storagebased on voltage threshold crossing. Cheetah data Wlesincorporate Xagged signals that represent critical taskevents (e.g., reward delivery) that are generated by Temposoftware (ReXective Computing, Inc.) which is responsiblefor selecting behavioral events such as trial type. Cheetah

Fig. 3. Helmet worn by freely moving monkeys. A Neuralynx headstage ismounted inside and to the rear of the helmet (10 cm long £ 12 cmwide £ 12 cm high) and is connected to a set of electrodes using a cable setbuilt in the laboratory. A lightweight cable connects the headstage with acommutator and ampliWers. The skirt at the helmet base prevents themonkey from damaging the electrodes which are chronically implanted inthe brain. The two holes at the base of the helmet accept bolts that attachthe helmet to the implant; the single hole at the top of the helmet allows acamera to image the head scene.

data Wles also store information about computed locationof the monkey’s head. Action potential waveforms are dis-criminated oZine with semi-automated and manual sortingsoftware obtained commercially (Plexon Inc., OZinesorter) or freeware written by ourselves and other investiga-tors (ElectrodeSorter, B-H. Kim; Mclust, D. Reddish)(available upon request). Graphical analysis of neuronalWring used NeuroExplorer (Plexon, Inc.) and VideoDe-coder (B-H. Kim).

Cheetah data Wles are relatively large; a 30 min recordingsession typically require 5 Mb for a hippocampal pyramidalcell with low Wring rates (75–400 total spikes), or in excessof 100 Mb of disk space for a neuron with a high Wring rate(hippocampal interneurons). While disk space for data stor-age is trivial given the low cost and large capacity of DVDrecorders, the time requires to analyse large Cheetah Wlesusing available software is a major limitation. For example,a 70 Mb Wle generated by a single tetrode could take 2 daysto discriminate action potential waveforms using a Dellcomputer with a Pentium III CPU and 512 Mb of RAM,even when the computer is dedicated to this analysis job.

2.2.5. Sample neurophysiological dataThe methods described above have enabled us to carry

out recordings in dorsolateral prefrontal cortex [18,19],putamen (Kim et al., in preparation), and hippocampus[11,8]. Monkey hippocampus, like that of the rat [15] con-tains neurons whose Wring rate is maximal when the mon-key walks through a particular region of the testinglaboratory [11]. Such neurons are described as having placeWelds: these cells Wre selectively when the monkey walksthrough a part of the laboratory but are inactive in walkingthrough other parts. In the monkey, the appearance of neu-ronal place Welds is not simply related to a location in thelaboratory as deWned by the presence of environmentalcues [8]. For example, neuronal activity (Fig. 4A) is veryweak in the Delayed Alternation task but a place Weld is

Fig. 4. Plot of hippocampal Wring rate in two behavioral tasks. In theDelayed Alternation task (both north and south versions) there is spo-radic Wring in the south-east quadrant. In contrast, a clear place Weldemerges in the north-west quadrant during performance of the 2 choicealternation task. Intensity of Wring rate is color-coded, each colored pixelrepresents neuronal activity.

206 N.L. Sun et al. / Methods 38 (2006) 202–209

evident in the north-west quadrant when the monkey alter-nates between the north and south boxes in the 2 choicealternation task (Fig. 4B). The selective appearance of theplace Welds occurs even though the monkey takes the samepaths and directions in both tasks. Thus, when the monkeyperforms a diVerent task, new place Welds can appear andold Welds can disappear; such changes are termed remap-ping. Remapping occurs even though the environmentalcues are constant. Our methods permit us to record fromsingle neurons for long periods of time [4] and thus it isunlikely that the remapping of the place Weld in twosequentially performed tasks is an artifact due to changes inthe recording isolation of a neuron.

2.3. Telemetric recordings of freely moving monkeys in a 50 m2 environment

To improve the ability to record in a spatially extendedenvironment that approximates that of monkeys in thewild, we realised that it would be necessary to adopt tele-metered recordings. It is unlikely that investigators inter-ested in primate social interactions would be able to obtainuseful data when cables are attached to the helmets of mon-keys as such cables would be quickly destroyed. However,cable-based recording systems are useful in the testing envi-ronment described above in which recordings take placefrom a single monkey as they can provide unparalleled datain terms of quality and quantity.

One problem with telemetered recordings is that theability to acquire data on multiple electrode channelssimultaneously has proved to be technically challenging.For example, Neuralynx Inc. has worked on a miniature 32channel telemetry system for some time, but currently moredevelopment is required before a reliable product is avail-able commercially. However, for investigators interested inacquiring data on a limited (e.g., 2) number of independentchannels, building a radio-telemetric system for freely mov-ing monkeys is readily attainable [9]. We refer the reader tothis paper for the design concepts and schematic diagramsof our ampliWer circuits.

The development of our recording system faced severalchallenges, for example, how to eliminate electrical artifactsgenerated by the rapid limb and body movements of freelymoving monkeys and how to provide information about theviewing perspective and eye position of the monkey. Our sys-tem is able to send neuronal signals from two electrodes aswell as video signals from two cameras mounted on the headand used for monitoring head direction and eye position. Forsimplicity, we chose to focus our eVorts on designing theampliWcation system for the neuronal signals but used a com-mercially available audio–visual transmitter for the actualtelemetry; the neuronal signals were sent on the audio chan-nels and the video channels were used for head and eye cam-eras. Monkeys are surprisingly strong for their size and wetook advantage of this by mounting several power suppliesfor the equipment, three ampliWer stages, and the telemetrysystem in a jacket worn by the monkey [9].

The elimination of electrical artifacts in the recordingsystem required a preampliWer with common mode rejec-tion; two power supplies for the ampliWers; and a groundwire from the second ampliWer to the skull. The neuronalrecordings employed diVerential recordings with a lowimpedance reference electrode located approximately 1 mmfrom the microelectrodes (0.5–1 Mohm). The bipolarrecording conWguration allowed a common mode of input,reducing the interference generated by large movement arti-facts which appeared on the recording and reference elec-trodes. Three ampliWcation stages were used with thepreampliWer and the two main ampliWers having isolatedpower supplies. The reason for separating the ampliWersand power supplies is to avoid feedback-driven oscillationsbetween successive ampliWcation stages as the output of asecond or third stage ampliWer could aVect the power sup-ply to the Wrst stage ampliWer. In addition, the power inputsof all operational ampliWers were bypassed with capacitorsto reduce oscillations; this is particularly important for thefront-end preampliWer.

A unity-gain diVerential preampliWer received inputsfrom the recording and reference electrodes and is used tomatch the impedances of the electrical signals from themicroelectrodes with the main ampliWers. Two furtherampliWcation stages were used, powered by 12 V batteries.The output of the second ampliWer (gainD50) is band-passWltered (900–6000 Hz) and fed to a third ampliWer(gainD100) for a total gain of 5000. The separation of thetwo main ampliWers is to avoid the feedback ripple problemdiscussed above.

Two video cameras were mounted on a helmet worn bythe monkey. A wireless video camera (JLT-1.2G, ShenZhen Company) is mounted above the eyes on the midlineand is directed forward in order to record the “head scene,”i.e., the view straight ahead of the monkey. A second cam-era (203CA, Shen Zhen Company) is used to record eyemovement and is mounted on a arm and directed towardthe pupil. Both cameras measured 26£ 26£ 16 mm. Thetwo video signals were sent to the audio–video transmitterin the jacket.

Signals from electrodes were passed to the audio channelof an audio–video transmitter, in parallel with the camerasignals which were sent on the video channels. These signalswere sent telemetrically to the data storage devices located50 m distant. The “data storage station” included two videoreceivers, a quad video processor, a videotape camcorder(held by the experimenter) and dc power supplies. The sta-tion is packaged inside a suitcase (26£16£20 cm high),weighed about 2.0 kg and is carried by the experimenter. Oneaudio–video receiver is dedicated to the signals of neuronalactivity and the video signals of the eye movements. Fig. 5Another receiver (JLT-1.2 G, Shen Zhen, China,11£5£2.2 cm) is used to receive the video signals from thehead scene camera. Video images from all three cameras werethen fed into a four channel black-white quad processor(Goldbeam, Qb-Bnc, 20£22£4.5cm high); the compositepicture of the three cameras as well as neuronal activity is

N.L. Sun et al. / Methods 38 (2006) 202–209 207

sent to a videotape camcorder (Sharp VL-E660U; or Pana-sonic NV-MX300), where it is stored on 8mm cassettestogether with the simultaneously recorded neuronalresponses. A rechargeable battery (12 V, 7 Ah) is used topower all devices in the station except for the camcorder.

2.4. Synchronisation of behavioral events with telemetered signals

One important problem that arises with telemetered datais the analysis and replication of the experimental condi-tions. Telemetry allows high quality recordings to be madewhen monkeys move within a large spatially extensive envi-ronment, as shown in Fig. 6. However, diYculties arise inanalyzing the data in a way that provides informationabout desirable experimental variables. Typical neuro-physiological experiments measure neuronal Wring ratewith respect to a triggering event such as the onset of avisual stimulus on a video monitor, or the arrival of themonkey at a particular feeder in a freely moving monkeyexperiment [19]. In the open Weld experiment with a 50 m2

testing environment [9] the variability of the visual environ-ment is a potential problem as the monkey can look atmany diVerent objects from many viewing perspectives.This problem is compounded because it may be diYcult toobtain repetitions of the stimulus events that activate indi-vidual neurons. Nevertheless, the potential advantages of

Fig. 5. AmpliWers mounted on printed circuit boards.

recording from freely moving monkeys in spatially exten-sive environments is very great—in principle, ethologistscould examine neuronal activity from a monkey living in asocial group, or neuropsychologists could record in freelymoving monkeys during the initiation and maintenance ofwalking in order to study neuronal activity in the basal gan-glia which would helpful for understanding their role inmovement and thus yield insights into the problems of Par-kinson’s disease [1].

The problem of synchronising behavioral events withtelemetered signals is solvable in many ways. For example,a simple solution requires the investigator to design a struc-tured series of events in which the monkey’s behavior canbe coded and a method (e.g., synchronised clocks in com-puters) for linking telemetered neuronal activity to thesecoded events. Behavioral studies of animals commonly usecomputers programmed to encode speciWc events (e.g.,vocalization, groom, etc.) via a keyboard in order to deter-mine the frequency of behaviors of interest to the research(e.g. [2]). Entering such events into the data stream contain-ing time-stamps of neuronal spikes is straightforward. Wehave used this technique to encode the onset and cessationof walking during recordings from the putamen and hippo-campus (Kim et al., in preparation) and found that it pro-duces reliable results for slowly occurring events such as theinitiation of walking from a sitting position.

The recordings obtained by the telemetry signal (threemonkeys; [9]) from hippocampus and parietal cortex weresatisfactory, comparable to those obtained by commer-cially available cabled systems we used in the laboratory(four monkeys; [19,8], and in preparation). As far as we areaware, we did not experience loss of signals at certain loca-tions within the recording area though this needs furtherresearch. The bandwidth of the telemetry system resulted inlow pass Wltering of action potentials but apart from somereduction in amplitude, the recording quality is acceptable(Fig. 6).

3. Concluding remarks

There are many obstacles to obtaining high qualityrecordings from single neurons in freely moving monkeys.For example, failure to appreciate the frustration of anunattended monkey can lead to the destruction of an

Fig. 6. Telemetered recordings of hippocampal neurons in freely moving monkeys walking in the garden of the Kunming Institute of Zoology. Waveform

208 N.L. Sun et al. / Methods 38 (2006) 202–209

expensive equipment. However, a little common sense canusually eliminate these problems. Moreover, the equipmentnecessary for the experiments are very similar to thatrequired in chair-housed monkeys. It is possible to buy vir-tually all the equipment from a commercial vendor, but inpractice many items are easily constructed in a competentneurophysiology laboratory. Accordingly, we imagine thatthere are substantial gains to be made into understandingbrain function in freely moving monkeys.

Acknowledgments

Support for the development of the methods describedin this paper was provided by NIH Grants MH58415 andNS-36255; Chinese Academy of Sciences Grant KSCX2-SW, National Basic Research Program of China Grant2005CB522803; Chinese National Science FoundationGrants 30470553 and 30530270, the Yunnan ScienceFoundation.; and Whitehall Foundation Grant A89-04;National Science Foundation Grants of China (NSFC)30530270 973 program project 2005CB522803 Chinese-Finish International Collaborative Project-neuro(NSFC).

Appendix A

Air Flyte Electronics56 New Hook RoadBayonne, NJ 07002, USA+1 201 436 2230

American Dynamics6795 Flanders DriveSan Diego, CA 92121, USA+1 800 507 6268http://www.americandynamics.net/

DragonXy Research & DevelopmentPO Box 507, Ridgeley, West Virginia 26753-0507, USA+1 304 738 3609http://www.dragonXyinc.com/

Goldbeam Electronics, Inc.1741 W. Rosecrans Avenue,Gardena, CA 90249, USA+1 800 777 4760

HYGENIC Perm Line Repair ResinAkron, OHNeuralynx, Inc2434 North Pantano Road Tucson, AZ 85715, USA+1 520 722 8144http://www.neuralynx.com/

Panasonic Corporation1 Panasonic WaySecaucus, NJ 07094, USA

+1 201 348 7000http://www.panasonic.com

Primate ProductsPO Box 620415Woodside, CA 94062,USA+1 650 529 0419http://www.primateproducts.com/

ReXective Computing, Inc.Sheldon HoVman917 Alanson DrSt. Louis, MO 63132, USA+1 314 993 6132 [email protected]

Sharp Electronics Corporation [USA]1300 Naperville Drive, Romeoville, IL 60446, USA+1 800 237 4277http://www.sharpusa.com/

SHENZHEN OMENT ELECTRONICS CO., LTD.5B Xingyunge, Zhongfu BuildingFumin Road, FutianShenzhen 518048, China+86 0755 83866791models JLT-1.2G (wireless) and 203CA

Smith & Nephew Wound ManagementPO Box 81101 Hessle RoadHULL HU3 2BN, UKTel: +44 (0)1482 225181http://www.smith-nephew.com/

References

[1] J.-P. Azulay, S. Mesure, B. Amblard, O. Blin, I. Sangla, J. Pouget,Brain 122 (1999) 111–120.

[2] S.A. Castner, P.S. Goldman-Rakic, Neuropsychopharmacology 20(1999) 10–28.

[3] E. Evarts, J. Neurophysiol. 29 (1966) 1011–1027.[4] P.A. Greenberg, F.A.W. Wilson, J. Neurophysiol. 92 (2004) 1042–

1055.[5] M. Habib, A. Sirigu, Cortex 23 (1987) 73–85.[6] D.A. Hill, Am. J. Primatol. 43 (1997) 305–322.[7] K. Hill, C. Boesch, J. Goodall, A. Pusey, J. Williams, R. Wrangham, J.

Hum. Evol. 40 (2001) 437–450.[8] B.-H. Kim, S.-L. Lim, J.-W. Ryou, F.A.W. Wilson, Soc. Neurosci.

Absts. 30 (2004) 1007.2.[9] Y. Lei, N. Sun, F.A.W. Wilson, X. Wang, N. Chen, J. Yang, Y. Peng, J.

Wang, S. Tian, M. Wang, Y. Miao, W. Xhu, H. Qi, Y.Y. Ma, J. Neuro-sci. Methods 135 (2004) 35–41.

[10] N. Ludvig, J.M. Botero, H.M. Tang, B. Gohil, J.G. Kral, J. Neurosci.Methods 106 (2001) 179–187.

[11] Y.-Y. Ma, J.-W. Ryou, B.-H. Kim, F.A.W. Wilson, Spatiallydirected movement and neuronal activity in freely moving mon-keys, in: S. Mori, D.G. Stuart, M. Wiesendanger (Eds.), BrainMechanisms for the Integration of Posture and Movement, Pro-gress in Brain Research, vol. 143, Elsevier, Amsterdam, 2003, pp.505–512.

N.L. Sun et al. / Methods 38 (2006) 202–209 209

[12] E.A. Maguire, N. Burgess, J.G. Donnett, R.S.J. Frackowiak, C.D.Frith, J. O’Keefe, Science 280 (1998) 921–924.

[13] E.A. Murray, M. Davidson, D. GaVan, D.S. Olton, S. Suomi, Exp.Brain Res. 74 (1989) 173–186.

[14] J. O’Keefe, J. Dostrovsky, Brain Res. 34 (1971) 171–175.[15] J. O’Keefe, Hippocampus 9 (1999) 352–364.[16] T. Ono, K. Nakamura, H. Nishijo, S. Eifuku, J. Neurophysiol. 70

(1993) 1516–1529.[17] E.T. Rolls, R.G. Robertson, P. Georges-Francois, Eur. J. Neurosci. 9

(1997) 1789–1794.[18] J.-W. Ryou, B.-H. Kim, F.A.W. Wilson, Soc. Neurosci. Absts. 30

(2004) 324.6.

[19] J.-W. Ryou, F.A.W. Wilson, Cogn. AVect Behav. Neurosci. 4 (2004)430–443.

[20] J.S. Stamm, The riddle of the monkey’s delayed-response deWcit hasbeen solved, in: E. Perecman (Ed.), The Frontal Lobes Revisited,IRBN Press, New York, 1987, pp. 73–89.

[21] A.F. Richmond, Primate diets, patterns and principles, in: Primate inNature, Freeman, New York, 1985, pp. 163–183.

[22] F.A.W. Wilson, E.T. Rolls, J. Neurosci. 10 (1990) 1254–1267.[23] F.A.W. Wilson, Y.-Y. Ma, P.A. Greenberg, J.-W. Ryou, B.-H. Kim, J.

Neurosci. Methods 27 (2003) 49–61.[24] F.A.W. Wilson, B.-H. Kim, J.-W. Ryou, Y.-Y. Ma, Behav. Res. Meth-

ods 37 (2005) 368–372.