A.system for biofeedback conditioning of electroencephalographic activity* A. T. Pryzbylik
R. C. H o w e
Department of Neurophysiology, Walter Reed Army Institute of Research, Walter Reed Army Medical Center, Washington, DC 20012, USA
Department of Physiology and Bioengineering, Eastern Virginia Medical School, 358 Mowbray Arch, Norfolk, VA 23507, USA
Abstract--A system is described for instrumental conditioning of e.e.g, activity in the awake cat. The system was utilised to experimentally control the state of voluntary motor suppression by biofeedback conditioning of a specific e.e.g, pattern previously correlated with inhibition of motor activity. This e.e.g, correlate has been termed the sensorimotor rhythm (s.m.r). With minimum modification, the system can be used to condition other Iocalised e.e.g. phenomena. Keywords--Biofeedback, Autoconditioning, Electroencephalogram,
Introduction PREVIOUS STUDIES have shown that autonomic (CHASEand HARPER, 1971), somatomotor and electrocortical activity (WvRWICKA and STERMAN, 1968) can be successfully regulated through instrumental conditioning. The system to be described was designed for instrumental conditioning of an electroencephalographic (e.e.g.) rhythm (12-14 Hz) recorded from the somatosensory cortex of an awake cat (Howe and STERMAN,1972). This activity has been termed the sensorimotor rhythm (s.m.r.) (RoTH et al., 1967). It is observed in the quiet, alert animal and is correlated behaviourally with a suppression of movement (Howe and STERMAN,1972). By utilising the system to reinforce the s.m.r, of a naive hungry cat, a conditioned e.e.g, response associated with stereotyped motionless postures will develop (WYRWICKAand STERMAN,1968). This type of training makes it possible to increase significantly both the occurrence of s.m.r, and the related suppression of movement (STERMANet al., 1970). The general operation of the system and method for instrumental conditioning of s.m.r, utilising biofeedback techniques will first be discussed. The system will then be described in terms of individual units and their function.
Surgical methods All animals were surgically prepared for chronic investigation under sodium pentobarbital anaesthesia (16 mg/kg). Small stainless-steel jeweller's screws were used for all cortical electrodes. All leads were soldered to a 20-pin miniature female connector and permanently mounted on the skull with dental cement. A minimum two-week postoperative re" First received 16th Apri/ and in final form 4th June 1974
Medical and Biological Engineering o
July 1975
Instrumentation, Behaviour
covery period was allowed before initiation of the study. Further surgical details have been described previously (HOWE and STERMAN,1972; 1973).
General description of system Fig. 1 illustrates the complete system as a block diagram. The basic system included a modified experimental beh~vioural chamber, e.e.g, recorder and a biofeedback device. Behaviour of the animal was observed with a closed-circuit television system. Table 1 lists the principal items of commercial equipment utilised in the system. Accessory items associated with the e.e.g, recorder included a pushbutton electrode selector panel, 'pin' jack electrode board, slip-ring assembly and feedermechanism electronics. Instrumental conditioning of the s.m.r, was accomplished by placing the animal in the test chamber and connecting the subject to the system with a 20-1ead shielded cable. Attached to the cable was a 20-pin male connector which mated with the female connector surgically attached to the head of the animal. The cable passed through an opening at the top of the chamber and established electrical continuity through a counterweighted slip-ring assembly. Each lead of the cable then terminated into an electrode board by means of a 'pin' plug. The cable and counterweighted slip-ring assembly allowed the animal unencumbered freedom of movement. The electrode board functioned as an interface between the animal and electrode selector panel mounted on the e.e.g, recorder. The electrode panel, consisting of pushbutton switches, was used to select the desired pair of recording electrodes for each channel on the e.e.g, recorder. All e.e.g. recordings were bipolar, although monopolar recordings are possible if a suitable indifferent reference electrode is available.
577
The biofeedback device, an essential c o m p o n e n t of the system, was used to analyse three preselected frequency ranges by active filtering. In addition, it contained a feedback logic control channel. Input to
this unit was an amplified e.e.g, signal from the preamplifier of the recorder. The outputs were a filtered signal of the input and a relay contact closure reflecting a selected amplitude level of that filter.
o COUNTER ~////////~
(PREAMPS) ILl
EEG
21
EEG
1
EEG
1
EOG
0
(PENAMPS) INPUT--OI OUT~T~O~
[ I
0 0
~0
BIOFEEDBACK DEVICE FILTER (5-7H:7) INPUT-(~ FILTER OUT-(Z) RELAY OUT-C) FILTER (8-1OHm) -0 --
o
SELECTOI; 9R~NEL
l
9
~ RELAY(8-10H~C~
I
~ELAY(12-14HZ)0 0 FEEDER 0 MARKER 0
I
FILTER(12-14HZ)
~/ I
RELAYLOGIC __ ---~DISPLAY
-0 0 -0--~---<3 O--
INPUT--C OUTPUT~
9
Fig. 1 System diagram for biofeedback conditioning e.e.g. Table I. Principal items of commercial equipment utilised in the system
Item
Model
Manufacturer
Behavioural chamber (modified) E.E.G. recorder Electrode selector panel Electrode board E.E.G. amplifier unit Biofeedback unit Closed-circuit television system Electromechanical counter Slip ring Miniature connectors
Model 133-10 Model 78 Model 67ES258 Model SEBN25 Model 7P511 Neuroanalyzer 3001 Model MTC-12 Model 1567 CAY-110-20 SRE 20PJTCH (Male) SRE 20SJ (Female) Part 250-3838 Part V52DA1100
Lehigh Valley Electronics Grass Instruments Grass Instruments Grass Instruments Grass Instruments Neuro Feedback Instruments Concord Hecon Airflyte Electronics Winchester
Cable Solenoid valve 578
Microdot Incorporated Skinner Electric Valve Division
Medical and Biological Engineering
July 1975
chamber. This recording cubicle had constant light, temperature and sound conditions provided by an incandescent d.c. light source, a baffled air intake and exhaust blower, and a white-noise generator. The walls of the chamber were constructed of solid steel plates which provided excellent shielding of the subject from outside electrical interference, such as 60 Hz and r.f. noise. A closed-circuit television camera was installed on the door of the test chamber for behavioural observation of the animal on an adjacent monitor. Contained in the chamber was a liquid-feeder apparatus which consisted of a liquid solenoid valve attached to a feeder cup. The solenoid valve was activated manually by a hand-held pushbutton switch or automatically by the feedback logic output of the biofeedback device. A liquid reservoir for the gravity feeder was located on top of the chamber.
Frequency ranges of the filtered outputs could be selected with interchangeable, preset filter cards. When the system was used to condition the animal automatically, one of the filter relay outputs was selected as the input for the feedback logic control channel. When this input satisfied the duration requirement as determined by the duration selector switch, the logic channel was activated. This resulted in a relay contact closure that initiated the timing circuit for the feeder mechanism. The contact closure was also displayed on the recorder in the form of a pulse. When activated, the feeder injected liquid food into a cup located in the behavioural chamber. The system also had a manual mode of operation. In this mode the logic circuit of the biofeedback device was bypassed and a hand-held pushbutton switch was used to activate the feeder mechanism. The manual mode was used for the initial behavioural shaping procedures during which the animals learned to obtain liquid food from the delivery cup. Fig. 2 is a sample record of a typical s.m.r. conditioning session. Electrical activity from the right pre-postcruciate gyrus leads in channel 1 (sensorimotor cortex in cat) was processed through the 12-14 Hz filter and its relay output was displayed in channel 7. Simultaneously, e.e.g, signals from the left posterior marginal gyrus in channel 3 (visual cortex in cat) were passed through the 8-12 Hz filter, resulting in the relay output in channel 6. F o r this session the 12-14 Hz output was selected as input for the feedback logic channel. Each time the filter relay output satisfied the predetermined requirements of the feedback logic channel, the feeder mechanism was activated. This action was noted on channel 8 in the form of a marker pulse from the feeder display output of the biofeedback unit.
Recording cable and slip-ring assembly A 20-lead cable was constructed from low-noise coaxial wire. Each lead in the cable was individually shielded to reduce movement and stimulus artefact between leads. One end of the cable terminated with a 20-pin miniature male connector which then mated with a female connector of the same type attached to the head of the experimental animal. Each lead at the opposite end of the cable was soldered to a 'pin' plug which mated with the 'pin' jacks of the electrode board. To permit the animal full freedom of movement and eliminate twisting or tangling of the sensitive leads, the cable was suspended from a counterweighted, slip-ring assembly. The enclosed capsuletype slip ring and brush block contained 20 circuits. All rings and brushes were of solid gold to prevent tarnishing and therefore ensure a high conductivity and an extremely low contact noise. (A detailed description of the cable and slip-ring assembly is in preparation.)
Modular description of system Experimental behavioural chamber Physiological data were recorded from the animal in a modified triple-walled experimental behavioural
SMR CONDITIONING
Slim
\
Pulse
,
~
,
8-1ocPsR~,oy "
--12 14 CPS Reioy-Feeder
- - ~ -
t. ~
~
- ~ - - - ~ - -
j,
~--
~
~
.... ,
,
~
~
-
~
[ . . . . . . . . .
. . . . . . . .
- - - ~ -
i
-
. . . . .
~
~_ ~i
~
~
~
s ---
~
~
-
-
~
~
J
~
~
~ [
-
fl, "
= 2sec
50pV
Fig. 2 Sample e.e.g, record of a typical s.m.r, conditioning session
Medical and Biological Engineering
July 1975
579
E.E.G. recorder and accessory units
E.E.G. data and relay output displays from the biofeedback device were recorded on an 8-channel e.e.g, recorder. Each channel contained a combined solid-state preamplifier and pen-driver amplifier in one unit. The input to the preamplifier section was differential and, hence, required two input leads to each channel. All e.e.g, recordings were taken with the one-half amplitude low-frequency filter setting at 3 Hz, the high-frequency filter at 0.1 kHz and the pen high-frequency filter at 30 H7. The paper speed was 15 mm/s. Relay display signals from the biofeedback device connected directly into the driver amplifier section. The numeric electrode selector panel is an accessory unit to the e.e.g, recorder. Pushbuttons on the panel select electrode pairs to be recorded on the various channels. There are 23 active positions for each channel. Calibration signals and test jacks are also available on the panel.
A numeric electrode board was used in conjunction with the numeric pushbutton electrode panel as an interface between the experimental animal and e.e.g, recorder. The electrode board also had 23 numbered 'pin' jacks, in addition to an earth jack. Biofeedback device
The biofeedback device contained three selectable active-filter channels. In this system, the three filters were 5-7 Hz, 10-12 Hz and 12-14 Hz. Each channel had a calibrated precision input attenuator, a filter output with buffer amplifier, and amplitude-duration sensing electronics with relay output. The input signal for each channel, after passing through the input attenuator, was processed by an active filter which amplified the desired frequency and attenuated all others. The buffered output of the filter allowed the filtered signal to be displayed on the e.e.g. recorder. Simultaneously, the filtered output was used as the input to the amplitude sensing electronics which operated a relay output when the amplitude
I MAGNECRAFT IREED W-- 1 0 5 - X - 9 (RELAY 4)
('F?
KHUI7D11 (RELAY2) +24 v "---00---~ SIGMA16000S I,C
I
5/~f
B
( RELAY I ) I
1K TIMING CIRCUIT
,.4005
"B K_I-
I __--
+24v "
MANUAL l 0 O'-S~___~AUTO
$5
,.u,,0,,
(RELAY 5) +24v
,TIMED
oo-i CONT,NUOUS1-O O
~ COUNTER
_2_ T O
RELAYCONTACTOUTPUT FROM > BIOFEEDBACK DEVICE
-~
FEEDER
SOLENOID
__9_
Fig. 3 Circuit diagram of feeder timer and control unit
580
Medical and Biological Engineering
July 1975
of the input signal reached a preset value for that filter. Each relay had two outputs: an output jack for recorder display and another which internally connected to a logic input selector switch. In addition to the three filter channels, the unit had a feedback logic control channel. A 3-position selector switch allowed any one of three filter relay outputs to serve as input for the feedback logic control channel. Controls were provided for adjusting the feedback parameters: input duration, intertrial interval and output period. The input duration, variable from 0-5 to 6 s, set the minimum required time that the input filter relay must be activated for a feedback output to occur. The intertrial interval, adjustable from 2 to 30 s, determined the minimum time between two successive feedback outputs (in this case, between milk reinforcements). The output period, variable from 20 ms to 1 s, was the contact closure time for an isolated relay, which was used to initialise the timing circuit for the feeder mechanism. Feeder timer and control unit
Fig. 3 is a schematic of the electronic unit designed to operate the feeder mechanism and electromechani.cal counter. The unit provided a 24 V relay output which was used to energise the feeder solenoid and counter. The mode of operation of the feeder, either manual or automatic, was selected with switch S1. In the manual mode (switch S1 in the manual position and $2 in the timed position), closure of pushbutton switch $3 energised relay 1 through C1. The activation time of relay 1, as determined by C1 and relay coil, was sufficient for relay 2 to become energised, regardless of how long $3 was depressed. The normally closed contacts of relay 3 provided a latching voltage to maintain activation of relay 2. This allowed capacitor C2 of the unijunction transistor timing circuit to become charged through the series combination of R2 and R3. After a period determined by the time constant of C2, R2 and R3, a sufficient bias will be established at the emitter of the transistor. This will initiate conduction in the transistor and allow C2 to discharge through the coil of relay 4. Activation of relay 4 then causes activation of relay 3 and subsequent loss of the latching voltage in relay 2. Deactivation of relay 2 results in disruption of the 24 V output required to energise the feeder solenoid and counter. The 'on' time of the circuit, variable from 0'9 ms to 2.0 s, was controlled with potentiometer R2. This time determined the amount of liquid reinforcement delivered to the animal. The electromechanical counter was used to count the number of biofeedback responses. The same sequence of events occurred in the automatic mode of operation with the following exception. When switch S1 was in the AUTOposition, energising voltage for relay 1 was present only upon closure of relay contacts from the biofeedback unit. Medical and Biological Engineering
July 1975
The timing circuit is bypassed when switch $2 is in the continuous mode. If switch SI is also in the manual position, depressing the pushbutton switch causes continuous activation of the feeder solenoid. This switch arrangement was employed for cleaning the feeder apparatus after each training session. Summary
A system for instrumental conditioning of electroencephalographic activity in an awake subject has been described. The system employed various items of commercially available equipment. The devices utilised in the system were totally compatible. A minimum amount of modification was necessary for setting up the feeder mechanism and establishing the internal environment of the behavioural chamber. It was necessary to design and construct a slip-ring assembly and associated cables. A unit was also constructed which contained a timing and control circuit for the feeder mechanism. The overall design of the system permitted the recording of low-voltage (microvolt) e.e.g, signals from an awake, moving animal with little or no movement artefact. This greatly reduced 'false' triggering of the biofeedback device associated with large-amplitude movement artefacts. Because of the flexibility inherent within this system, its basic design could be utilised for biofeedback conditioning of other behavioural as well as physiological phenomena.
Acknowledgment--The authors wish to thank Mr. Lee Fallon, Department of Neurophysiology, Walter Reed Army Institute of Research, for his assistance in the construction of the feeder timer and control unit. Thanks are also due Mr. Sidney Ross, Neuropsychology Research, Veterans Administration Hospital, Sepulveda, California, for his consultation on the biofeedback device. In conducting the research described in this report, the investigators adhered to the 'Guide for Laboratory Animal Facilities and Care', as promulgated by the Committee On The Guide For Laboratory Animal Facilities and Care of the Institute of Laboratory Animal Resources, National Academy of Sciences-National Research Council.
References
CaASE, M. H. and HARPER,R. M. (1971) Somatomotor and visceromotor correlates of operantly conditioned 12-14c/sec sensorimotor cortical activity. Eleetroeneeph. Ctin. Neurophysiol. 31, 85-92. HowE, R. C. and STERMAN, M. B. (1972) Corticalsubcortical e.e.g, correlates of suppressed motor behavior during sleep and waking in the cat. Electro. enceph. Clin. Neurophysiol. 32, 681-695. HowE, R. C. and STERMAN,M. B. (1973) Somatosensory system evoked potentials during waking behavior and 581
sleep in the cat. Electroenceph. Clin. Neurophysiol. 34, 605-618. Rorn, S. R., SrERMAN,M. B. and CLEM~NTE,C. D. (1967) Comparison of e.e.g, correlates of reinforcement, internal inhibition and sleep. Electroenceph. Clin. NeurophysioL 23, 509-520. ST~RMAN, M. B., HOWE, R. C. and MACDONALD,L, R.
(1970) Facilitation of spindle-burst sleep by conditioning of electroencephalographic activity while awake. Science 167, 1146-1148. WWWfCKA, W. and STERMAN,M. B. (1968) Instrumental conditioning of sensorimotor cortex e.e.g, spindles in the waking cat. Physiology and Behavior 3, 703-707.
Un syst6me pour le conditionnement biordtroactif de I'activit6 electroencephalographique Sommaire---On d6crit un syst~me pour le conditionnement instrumental de l'activit6 e.e.g, chez le chat r6veill6. Le syst6me fut utilis6 pour contr61er exp6rimentalement l'6tat des suppressions motrices volontaires par le conditionnement bior6troactif d'un trac6 e.e.g, sp6cifique pr6alablement corr616 avec l'inhibition de l'activit6 motrice. Cette corr61ation de l'e.e.g, a 6t6 nomm6e rythme sensorimoteur (r.s.m.). Ce syst~me peut 6tre utilis6 pour conditionner d'autres ph6nom~nes e.e.g, localises avec le minimum de modifications.
System f/Jr biologische rfickkoppelung der E.E.G.-T~itigkeit Zusammenfassung---Es wird ein System fiir die instrumentale Konditionierung der e.e.g.-Aktivit~it bei der nicht betiiubten Katze beschrieben. Das System wurde verwendet, um den Zustand der freiwilligen Bewegungsunterdriickung durch eine biologiscbe Riickkoppelung eines spezifischen e.e.g.-Bildes, das zuvor mit der Hemmung der Bewegungstiitigkeit in Verbindung gebracht wurde, experimentell zu kontrollieren. Dieser Bezugsbereich des e.e.g, wurde sensorimotorischer Rhythmus (s.m.r.) gennant. Bei gerinfiigiger Modifikation kann das System zur Konditionierung anderer lokalisierter e.e.g.-Phiinomene angewandt werden.
582
Medical and Biological Engineering
July 1975