.0 I 4 1 .0 A 0 . .. 0 . I 4 4.0. : : .0: ‘0: :00 : 1 001054.410? 0‘. 40 0 I r .- .10:.1100.1.v.v00’9.01 10.n‘0NIN 3114.01.2011u1dfinu‘ 1.4.1.110...- 3.‘0:¢_.. . 0.. I.4.......000010..III ..0.. . 31.0: 11.4 .. .0 0.I 0.0. . . 04005.1...01. .. 150:” IF “I ‘0 411.009” ”..0 Id. .. (*1.::.«.&.1... 1 .02... .. .. 1. 1 .. 1. 0 0.1.11.9..un.....1.‘\fi_u.1.1m10‘n .1‘1 . o 00 .010: .0 1\ .. .: . . I . .01 N 10.....- :4 0 1 n. 100::0_I 0.0 :0. ...40 II ...-1.10;. .1.1.... 11.9 1.1.0. 9.. .11\1...0 ......110 .«1‘11~.0.. c.0\l..1 ..0 .o 01:: I.. 1. .00. 1... . 00 4. ..0 1 00.»: .00.: 0| 1...... .H.Ifl 1.1.1.1131... .:..11...1...10 1.1... 1.... .. 1.. .1. 1.1.1.1...0 Lick? .. R ....0 10. -...-. \.: 0.1-0 : .I..:1.... .. ..0-...}.- ...0 . 0.... 1.1.4 '0 . .10.. 0- 01.1. 4: 9 ..‘...I1..l‘.... D 1f05‘1”69..\1-.1l. 1.0.1.1 I. .I . 11 .‘..1.0J...I.O . . .. . . ‘4. 0 .40 II .8 0400 0J~4III :4... $4. ...10". 104010 1 .... . “A. .1. 71’19141..Ov. . .0(0.:9\10.13t..100.01 . 4: :R1:.. 1 :00. 0.0 : I: .1.0.10 0 00... -‘0 .0.r:0.l1 (.0100...- ‘0100‘ . to .0011‘01 ..J11I111111. .19.‘:1..». ..0 1. 1....010 .0. 1 1”. .0x _.1..00 . ...s... .:0 . 01.01:..14. . 4:. . . :.1.1...0.1 . .001. I ......I00uI :0 .0.. “I III, 44 4000:. 11‘. 01¢.I 4. .1 .1.1 14 5‘.” ‘09 ‘ {01. 1.0111100f:'.r~0.1.. \.. 010. 000:1030000‘ 1. 00 0001 0. .0....0 1:0: 4.0.. ...\.0r‘..00. I40. .100‘.‘.. '1 0hcv.F1.I1...:1.r.II. . 0. . .1 1. 01 1 .:.4 1. 0 V1. .1...1._1.II. .1 . .....VV . 0 .:r.,..... .0 0 n: - , 0... .10. .010-.0$1. .. 0.. Ifli-10a0 .10 4‘ n . ‘400 (0.1 1 . I; 1 a, 21?. . ‘9'... t.’ §1J.0..00..00.1 .... b01111.n.m1.{.'...10.\l.1 10.0. 1.... .. .r .:..0 00...1.:...1: 0.. -.. 01.011000 ...1 0 I 0... .4‘. .1...‘ Crud \1. I ..V. 1 1e 1 1 -. .Ifi’..0414.... . .. ..0.... ..1 01113.01 ..k‘? .- ‘4: . .... ... .0 1 .00. . 4 I :0‘.0 I .01. 44.01.10.0'1. 'It 01Vv0f044. 010 ..000» V0 517.“ 1’ 004.1 . v.1. 2.410. 00 «1... I. If. r0 0.0.1.001 11(00 0: ..0. 4 .I I 0' 0.01 I ‘0 00 III. .0 40.:1‘05. 0O0.40.0. I I .1... ..\ ’11.“. (N11... (I 1 0%..) .1101 : 11. 1. 1A 00.. ..0 1.110111. 4.... . I. ......01 101000 0.1- .. :00 .011 .. 0101.10. 000 I 0 4 00 9.0.; {‘04 01001 1§‘t'0'~104 . :0 .0410‘51. Q :U... ..rvhvr’0115N. ’oVL‘h. ‘.."A..11§ 6 1‘: -1 1‘00... :1 . 00 0.04 09‘. 0 00000.00. 0. 00.00 1 1.11.111 17.0041)? .0011011~1. .1 1.1? $0.00....” .1._.1I .1 1.9..1. ..110 01 ....... .1... 1.. ...... .00 10.... . 0.0010 ........00 .. .1140“ 1»MIII...1. ...H11. 7. . . L51 1 1 . 0.....014111 .1! \Ioe...10. 0.1......0.... .«0. 10 I. 1 I ....I: : . _ . 0 ..I:1....1..1.11 I 1.1% A .. 0: 101.....19. 3.104.101... 1 J 1:13.110 ..1 11... . . 1 :.1I....:I. .:.-115A 00... .... .1: ....:-.10.1.. 0.1. .. 4 0 ..0.. O 07.-..0 ‘S. 1 ...)... .10 4.161:..hr...'”01.1.v\1..11 . . Q.(....:II"I0. .0.‘. 10.0.... .0.000.: 0100 x 4.0. 0 .0 . :1 0r '1I0I0 10 410004h40. N . . . . ..J'U0O‘... c.1519'0’wb’usi‘vn.’ ,0). “04.0.4. #100 7'" ' .:. 1 .....t1141. 1111.....1'0. .0. 1.. I11. 0 ..00 .0 1 0 01 ..0. ... 01000‘. I 0 «I 1 t .0 -040. .11: 1 . '1 010 01\0. .:I:0 0 : .. .I. 1.. .1 .00 . .- 1.1..44 . I 04. 1 1 .: A II 11 0 .0. 0.. I1 .I : .00 .00. 0 41 . ...-.. ...r... . V... 1.3%.... .....I .....1 111...... ......n. .13.... 11.2.1.1 ...... .... :..:_....... ...... .... . ... ..I. ......YIQ. 1.....- .. 1.4.... .-.. ...... .:.. :. .-....C . ...»..1 I I 1. I. 0 W0. 1 4%.“. 10‘. 01.0...Y .4100‘. .... 1 1 0 'I’OI. \1 141...“ $0.01. 911‘0‘4_M,.1..01II0 1 0.x:‘t0. 0.1. 104,214. I.‘ I 4'. o .« .1.\0.Y0I0(. 14 1 . . ..11. .040’. 004 . 1 ofi“...0. .0.0....V.I:I 1:07)- '01.... '1. . :\.:0:: 114. 1.01. .11 1 10940 . v1...- 0101.1. 1. .I. ...I110 .. :00 :v' .. ‘11.»... I . .0 .I 1 110 .1 $1011 0 ..0. 10 . 0. '1. -14-. :4 I .0 s.\1 ...I I: 0 11...... 1. ...01 01. 0 r w.§.....,.4.:€.0. 1 .....WI.‘ .11.. . . 1. ......10. : 3.0.1.90: 09.21010. .‘100..~ on... I... -. . .0 ..0 . 01 0.... ..II:P0I.. .10-... .:.... _ L. 1 . .50... I .: 1.011. 1. 110.33% 5.1 . 2.1.1.403 .. .0 .... ...... 011191.: .1}... 0 . -.. .0.-.1 (II- \. 1.... ... :30 ..00. 1. . ..0.. ... II 000 01.1.0101- ‘11 . . .0 011. 00... 1.1101 0010 I‘ll. 10 001%0 0444 ’00 ~ 0001 Lvr. ~.O.41.4.0.4..0V1.$. \.0:.I4II 1 100.0 I... ..0: II 0.01 . L000 1 Q I 1 I. 1 - 00 400001 .’0110. 0414.10 100. 00. 9 . f 0‘: III4000 1qu*,. ...... 0*77«11.40 001-.0’0. % . 0 ... 121000 .0”.. .104. .I I k 4.1. '0 00 1 0410000 00 r. : .90: 1‘0 01004 . 91.. 0.0 11.0 .00.. I w 00‘ 1 0 .'.. 0.4 1‘0“. 1:”.0a1' 0 ...~1 .0 0'0 14" 100100 4.... 0 I1... 100: 0. 1... 0 I $11.01.".111: :.I.:11I1 I 0 :1 1.0..I0000. 0 10:! 1.4.0044 .: I1... 01.0.:H011 .9110 ovlefI ”0.0.‘9‘11 .. 1 0.101.011 1 .r' 10 .00. ....0 :J .00‘.-. 1 .01- 00.11.11. 1.1. 0. \ .0 I . :I..0: . 0004.4 44 1. 0......0 . 14: :ler 1..0)0 .1. 00 .00 1... : “K034131131... .....I..110: .....0040100.1.I. ......100. .0. . ..0....1..- ... 0...... .:. $00.11.- 0*. .1. -1 I. 0 . .. .. ..- .u .11.... 1.0.0.00 ..%.Ia.....I-.l.1110 401..109.0.)...01.. 1011u1.... 01.00“... 0&310.0.. 1. 0.0554010111113-::I1.11111..1\1\1$1.0...111. .:I 0.401- ..0.0\...I.101 0.....1.:0-..I... .. ..0.:1 ..0 .1 .0 . . 201.1... . .c 1 10 40.01: . .1 . .. - o . .... 0 .. . 1 4 1 01.... 9 .100 . . . ...... . {1.1 . 0.0 1. .....: 1. .0 ..0. -..0 .:JII ..: .. ‘1‘ . . . . -.0 .I 1 01.1.0} 11000.0...‘1030 : 1. ...OIP.I0..0..N .1 1...... I 1 41.0.4.11AIJ $.31. 0. :1...‘..... .1010.” 1110.0 .3111: 11... 4.0.04. \0:. 1\I1.0~ .. 50:00 . $10 . . I. 4 : . : 4 .1 Y. . 0 I . .NJ . .0110": .1 0 ... 4 .1110 00.. IN .18. - .04 .1! u. . 0.9.01.21 112'... On 1.. 0.1 .....kde ..0.... ..I00.. ..1.:.10.1 1... 0: 0.0. - ..0. . .....0110..r1 :....1 w: 00.01.100.72 . .. Ii.......0... .l:0..1\.. . . : Y: .1 1. -. .01”... .....0...1.: 1 I k 0:. . 31.001.030.12. 1.11.:00“: .1... .. (1.0.1.100 00 III .1. .- .0 .1000... .:.0...... .0. . . :0 0 I .0.: o. -:..0: . .100.:\I.0L00f-. .0 .0 : .. . . . 1 I .0: 1 . I .1 -. 0 .: .0. . 1.. \0 1. 0 . . . 0.1.4 4. . :1 00011 . .0~ . 1‘1 .. I 1 v 1. .. ...0.”1104(I.. .1... 1. I.../. 411‘...1140......1..Is000100..0101..102.40 Win"... .001 0.01.3.1... 1.1.1.0....00‘01..400 .101“ 10.0:01.0.401... 4154044.”... .::.04 0.01 : I. 01. (0 I. 1:! 1. I 1 .. 0 11:01 I ..10. . ....0111.II. . .1. 1 1.1: 1...... w. 1.: ..1111101 ...1. 1.010000.01.1.1.. 1 l. 19 1.. ..II 00.. . 10.....1... .0.: ..0.. ,0- . -... ..:_.:... 1..- I. 3.0.0". r.00.:'0.:1u0 070L4v010o401.0'00. ”0910 .44.)‘..01 J‘J’ I2».h0'0.‘001:'4.1 0. ..00! 044041... 4 0’6‘. 401.. .1‘.. II .0 ‘10.:\ . ‘40. A .714. 0... .l' 0:04 :I I .0 .: :V .: :0 . .. 00:. I 0 0404 ..6. ..45.4 ...10.. 001.010 .1 1-0. ... .: . 111145.17..-. 0. fr“. 5.0 I ...... 100.. . 0.0. 1. 10.1 ...:1...: .1 .:9 .0 1... .1 .1. .... .. .11 .0 .I I0 0001.0 00. 0.0 I: '0“... 0 -.1 I 140. :PI‘. 904.1.4V1011'00600111 ‘.4'0..: 1 11. 1 01.04J01¢11 1 0 ‘0. I .1’11040 H1 . 1 0 . 1. 00-404 .0....00 1 01.1 . 0% I 1. :. .0104. ...!If. ....)-I. .....u :Fflfio. . .0... ...: :Al . I. 0.01011 11.7040... 01. ...0. 1 . . I 1 “:01. 15:. 3-... 10. ..0.-0 0 ..0 01-1.30... .. ......0 0 . ...... I: 1.4 . .1 .0.. .:I... f . 0.1.0000» ..0.. .1 010111.00 01.10 1.111... 1.0! .01 ... I. . (.... ..II..1...01..0....1..70. 0.. \-. ... I10... .... .0.. . .. v .1. . 1 . ..0 .100: o .. I : . 7..- . H -..l. .0. _ ..0.! .I . 1:0 t.”.:. V 10.4 01..I.0:.w. 10 1 . ‘\1III.4 0 ..1-1W0000.10:.1..0HU.._W:‘O11910-109. .10001 0~._v..0¢0.1 Q . 1.01.0411 ..40 04.11. 00:04.10"... .. 00 4. .:l 01:44.40-:I .0001: 1.14.. 44 d0foI.4.040II 0.01: 0y... 0'01..\101 1.‘.u1011.-..IIH41H0«.."0. . 0.51:1 $00.11....0111 11.0.1. .0! .0 II ...0 7:4..410 5.100..I-0 I .1. .0 : ... 41:00:: :- _ '0 .PV 4. 00“. ..1Ol4:00‘1.:0:m0 .:I04I>I'~ ‘0'. .10... r I 0.0...v0’00rr 1.0.00.0“ 4. 0‘ 00 1 1‘ 00.41.0011. .:‘I .. 10.1 .1 .1 .01041.~I00 0119.13.11.01 . - 4 00. 1 0400.0 4 ’41 0'14... 1&1. .1 1 ..0000|:.. ... 0.0 ..10.100 .00... 00:. 1:10. ‘30...0. 00. I 411 04.. 1.11... . .0 0 ..0.04:0(1n100 001 H 0.. ..0-.-0.. 1. . I... ..0 :0 09: I :..40 . 1‘.J I I... .1... 00 01 L110. 1 . 100 40 . 003000 0100.. ‘1“01‘. 0.0.0..000~0.1 30.0.0.1v40. .41....I100090190 1:00... 0.110. 1 ‘0... 1 .00 0.0: ..0 .0100I 1. 10. .I: I 0 0 0 1 1 :I 00 1 11 0 1 00. .0 0 1 0.. 00. 4O. . : '0‘. 00. I 040 4 -0 0:1, 0 .04. . ILVFDOQ 1 04 .:0' 10:01.4'4I.'11 1.1. 1 .:.... ‘o .1 1.01 1. ..,.:1.. 00000. IIfl.‘ § I O -..I I. : 0 C I 0 0 I . 1 I- 1.4:. 0’ .010104... I v0.0.r.0.0:'. 1 III. :40. i 01. .0 4fi4010 4 I 4.0..-. .:0IJI.1...20..011410041:1Y9I 0 .Ii...§0 101.1. 00 1 ..1 1.111"...\.. ’4 .II‘. OOIII. .:.! ..0. - 1.0 0 . 1 0 .00 . .0. (”‘10 I ..0.?0.....I II..I00:I 1.0...00 1.: ....1.... 0.401.. .0.00 .0.... 30.0... . . 010001... 1. . .... .v .1 ...4. .... .0 . ..0. 1.1. . .0.-..r. . 01$... . . . .. : . 0:00 0:.W.: I 0. 01. : I . 10010 :0: ..101. .0... 0.. I? I... '1 ....0011. .0.01001.II......I 0... ....-.(000 1. .3.......1...r.0.0. 0.40.. ...00. I :. -0.01.: -.1 1. 00% .:.. .501 .0 .. u.\.0~..0-. :-:1 . .I .I - 10 00.000. .1 . ..1A .1!..1.0II1fi100..0 0’0 . 141...; [01.00 1.011\ .I .....1\.:.. .. . 4(010.1....:1:...I. 101 . 40.0111... ..0: I: .1. 0 ..- 1. 1 011 1.Y11_...01.I..0.01 . 0.040 ‘ .:0. I. 1 l' 000. 010.00.11.01 .1‘0.IP 0.10.1010.1.11$.0.I.1:.1\.0101.II\00.0010 I 2.. ...0101..Y1001_00b\ 0 0. II. .000.0 .0.h10:1|...1\ . . 4 ‘10 .O 00.1010 - 0010.10.4'xu4. I 4000 I. _ . ...:001 0 .0. ....-.1. “......II O0.. . .1 . ..0 1 .. .010 11.11.11 I: .1: I .. .I ..0.. 10 $1 : .IIYII. ., .MIII . 1 I. . :.011010..00 100:0..01.T-.1 - 0000,1011Q\40 10:-I...10-I::Io . ..0. II... .0030... .‘4'0 00.0.1.0 01.041310 0.....I.0 .Q. .1.0 : 0 00 0. I: . .0100 1 .. 0.1.1100 .0 (I... :.5.. .000 .000. .70 ::10 00 I040 01 4 41.40.00 4 l O. .04 1.400.104. .h 1 0 . 04. . 0....» A.“ ... 10.1. Q 0.0.4 ..0“? 001 ‘04 .. :0 00110 0 :d:0.1. )1 .II 1 . 0 000 0 J . O 4 0 00.240 01 0.0.4010.. . 0 .301... . . .0.. 1.10..1..W.0.0..0.0:110.101-...I. .I. .u..000: 15011.... .1 0..0.v.. . wv I: -... If....... .30. .0111... . ...-I... H0}; ..0 0... 504 . .:. .1 . 0 40:10:10. .0.-4.. ...0101... . ...414...r10.410 0.1 ..: ..0.1:..l ....4”.F.01.I..:0..\.Y 1 Jr. .. (1.110.? I 4. ..0‘1. 1. I I . 10 . h 0:1"..00 0).. r :0 0.0:..- 1. 1 0 . 10 04'. ... ”01.0.0. ..1 . .... 00.04.10. .1 .... .......1-..... . . 1 01101.... .1 .0. ...4. ... . 0 . . .:.. ... .1 0.. {...-k. .:.. :01 11‘0111 I . 1 4 041 140 100::0 011,400 0 0.0 000. D 1.~0’.. ..1:...:.. ..4' ‘1 .r01 1040 .000...'1.01.I.1 1. .4. 1.110. 0.: 0 0 000 . .014 . 0 ..:4.40.0 1001.. 1. d0. . 0.0.0 .0: :. 0.0 10 . 4. \100 I . 0.0.01 0.! 0 I' 0.\..01‘0II . . .. . .':4:010 0 00.40.1000.... .(11000. . to. 1 .1I'Q 100 .0 II. ..0.. 0.0”. 1 10.10 0&10. I : I 0 1 .0 0‘. 00.4.0..\. I 00.010 0 10 0. 0 10 .1... 0 .0: a...“ 1:4. . .401 04. 0: a 1 0 0 1 III: I 0.1 :0 ” «”00 0‘: ..1 I. : . .: . .0.- ...-1....00111... 7.1.1.419.‘ «0 .14.. 1.1«11:If. 00 a ‘0 0001. ..0. .I .10 . 1 4'1. 1 01: .0, .4.N0.1I 6 v.4 0.4 000.....- . . '01. *0 1000 I104: .. .:10 1:000 .. 1 1 0.1 0.4.0 00 10.44011. . 1.1.0 ‘40III'I.I.:I\..0. 1.. I). .4 01140040 ...--.. .4. . 004 :...:III '11 I . 0 .I 4. . 01.0.. :50:0:. 00' 0:11... '0 .1. .0 ‘44. . . I . .0:0 1 . I410 .. 4.0. «.5141. . :. I . . . . .. . 0. ...! .0 0.00.01... 1A0. 41.1.11. 1.0.0.. .00 1.1. 04.4.. .... .d 0.40 ..0.. : .I‘ ..0 . .0. .0. 1. .:.. .:.: (90400. 0.. 0 0:10.- 0 I:0|0 :I.I-.\.‘0. :0. 04.40.“.4. .0010‘44 . I". 00’01'0 .100 "001‘ 1 1:4. 41.10 . 1". 01000. :4. 1.0 Y 1 4- 0.00 .4 4.04 00:1 0.04 000 01.1.0“:000| 00. 0.440 .4 .0 . 4 10. ‘4 .0 00 0 I. . .IFI . 0 «V0.40..4413:.:10401 11.1“. ‘I 11‘... 'QII 1 \I l 0‘. ..0 I :44 4. 0 0:: 00\‘101000 I 1: .I . 4' 0. $0M... . .04.I..4 I 4. 4.10 .00. 00.01 :0 Q4404 ..4 00:.404 .. 4 IIQ*QOOC 1.41. 40......- .§.V'\I0. 1.\ .0000. IV .001.0Y:0.\0I 0 ’1 \000 .00.00.11. - 1.1 . 4 0.41.10. .. 10:011..-.400 1.010 0 . 0 1‘0‘0 D400 0.8.11.0.. . 1’0...IO..V:I ..0.!000011”... -001... . .1 I I :0 .: .0 I... .. r000 0.0!.0. .1 4~.:A 0.0.. 1 0. 11 I00 .:.... . . . I. 0 0 0.0. .10.. .:.:....l..00 ...1Y... .:.. .. .....00:. . I\0. .. :00 ... I ... . 00 . .0.-10:. : .1110... :0. 00.01004.HH .0.0 "00 ..I.. .0. . .. . . . . .010 14. .. 0 .I 14010.0...014:01.1.II:00 . 010040.? 0: ..0.. 04041 . ... . 0.A.. “. L110 -10401 .01. :00 1. . . ... . . 0. 10.!0... : 010: 11.11... . 01...... ..... 0:10 ...-4. .. . .. . I II 4:00:40. 00 00,0 4 x. s . .I. . .: 4:001: I .0... .- .‘00414 1 .1 . 0 I 00 :0 d.‘r.L‘f.I0. . 0 ‘4... 4.. I . : 01.1 0 . II. .0 .01.1r:l’(11:..0:. ...1..1\:..1h..u:. 0 .414. I. 44.400000: 9. :.:0.l..11... 0... . II 1 .0 ...: . .. .0. 0 H:- 4:. ... 0...»..1. . :-4.:111 . 0 .11. 000010040. 1 0 : I 4 0 04 .0 4014 1 .110 Y. 1 I... .: 1000. .: .071 .700101...0..10140.....4.4. .1. .00. .000 00 0 . 1. I:.. II 400.... .000: 0.1.1 .04 1 1 (0” 00 00.:1 I0. «I. . 00:90 ..0.-I4 V1-00 I 400 . 00 4 . . . .0 . V0 . . .0... .1 0..040.1._.II:..4 :00..10y....00,001.01.0 I I. 0 :0. '.11\0..00Y.I4 . ... II....IIII0I‘. ’ 404 I0 0. . 4 .0‘104‘ .:I'1I10.440.40.I 0 I: 0. -.4':0....(4 4 0 1 I 0 I 0 : 0 I v I f I 0 I 0 I y 0 I 0 gal I~:' .. - _.__. ' ._ .. 0.' .i I ‘ I D 0.. I: 1." 0 P .O I" . 0 r 0 l'. I I .0 ' 0 00 I 4 1 I.“ I 0 .1 ' I: I -',. .I 1 4 ...A L. 1 : - . a 0'1 1 0.0400 0 . 0 0 I 0 0.0 00 I. 0 I: 0. :00.40.Tl 00:1 0. . . 0 4. 1:. : 0001 .... 01:11 I 1 . I .0 .: 10411.0 .. 4 0 .0 . 1. 0 0 . 9 00. 0 1.1.0 I 0 I0 0 '04 0 1 4. 400 .I I 0|.:: 0 ... . 04 4 :. - -...1 0 I0. I I 00 0.. 1 : :01 I11 0 . 1.. G- 4 I 0- I: I . 0.. 4.4 I . 0 I‘ I . 1 00 0.0 ..0.-00 0 0 4 40'. 0 . 1 I 01 .:0.0.. 0‘ 1 : 1.. 0 I: . . 0.0001. . . 0: 0 . . .:. . 01 . IIIOR: 040' I . . . ... 4. I 0 1 .: o 0 ..- 004 ....II : - i 00 .. .:..0.0I0:.... . ..0. 0. .-.. .. . . . . . . 0.. .. . u . . . :: I:.I|.::..l...:'..... . . 4-46“... {4.0. in." :0'0115 '.. . .. . . 1 I. 0 0 :. 4 . . 00.: . 1 0 \I:0:::.$4.0.. .040 00 . . . I . ..0 1 :11 11 I .. 0f 0 I0 4 0.3.14 :0: .:.0 4 {40 II 0 40 : .1. : I: 040400-000 .0 4r. 1 00 0:0 I ‘l 0 1 . . 1| 0... . 04001 ..:0 1: 0. 1 00 I 10 I I I 44 4040 ll 0 40:0 0:0400000 I 0 . :Qr‘. 00 0.0.0010 00 - 0100 01 0 0 40 I. I :0 :0011 4.0 I 00 10410-0000I1441V4rlh1- . . 110040 I . 0 : 14 . J. . I ‘0 I . 0.0 I. 0 ..1 .:0 . .-0 .. : .. .0 II III. .I . .I. 11.0 0000.0:- . . .l...0: 00”.: 0 000 - 4.1.1:.1I.:. .1. .000: .....4:.00.0.4:.1...::::0... V 00....0.4 0'. 11!... 0. .7100 I...a .00: :1 0 .l. 0.... 1 .00001rltudl00 .... 0.1.0 .IWID 1 ..0 . 1 I . 0 I. . I .. . -00 0 . II .. . . I 4 : : 000 I 1. . 1 1 01 1.4 1 .0 ,0. . I 4 1 I . .0 . . 11 HQ. .0 I... . 0: I :0 0. .0. I00 0. ..0. .0 . I 0-...0I0.. . ..- .40.0‘ .01: . '1 0. 0:0: 0.:.1. 00 .. : I... 0 .0 . .040 :0 . I. . . ..01 I 1 010:4. 00 . 1.4.:00 00.. 400W 4.0!! I: 0 0 . I I : . ...0.‘ I. I .:. 0:0. I... 00 004 [I 4 ~ . 1 10 .\ 0::. I. .0 . . .1. .. .-0.. . .1 1 0 1.1:0 . .00 . .. .3 I . 0 ... 1 ... -0. . . .:.“1: I: :I .:.:l 00'... I.1.:. :.00..l:(1\ . .1 I I D I 4 I 00000 .00 II .4? .:....0 4 004.00.:0 0" 0000 I 10:0000 I. 0 4 4.0010 0 0 4104 0‘. I0 :11: I J1r0 0:00 : ‘1.) . . . . . 'I .0 . . 0.: . I : o I I . . I0 I0 I 0 . . v.0400:..\&4.00000.04—r .. 1:1 I 0 III“ 141 If .00 0‘11 V 0:. 00:1 1 : O :I..0‘ 00 1.. I I-I I. 00.. 0400 00 1|. 4: I 40.0: 00. 0.1 0 Y. I 04 01:040. ...I 0 \I.1:II :0OVI04I..000000 040. 41.000101‘0 .1. . 0 . I. .. . 0.. 0.... . 1 .0 ...: . I . . . . . .0 .0 I I: I .00 0 . fi 00 . 1 . 1'. .I0.101 : : ..0 0: . .:II . 0.: 10... . 0 : :0 000 L 00.012.011.004 0. II'L . 001 0.0 1.1. .0..- 1 I. : . . . 00... .0 .I .001: 01. 004. -.0 :0 . 1 0: II :o 0 . I 00:. 4 0 .. : - .I I c. :0 ll" 0:. .4? 4 . . a I . 404. o . . . . ..-III :0 -. o. ... . . .0. . . I . - .. 0. -... 1 . . . . I. ‘0 I400 - ‘0. .I Q . .011 .0 0. 0.0 4. ..0 0 .0... ..- .: .0.. J...1 I . v.1 1 1 . 0 000:0 I: 00. ll I‘|.: 1.0. 0.00.. -0 l 0 : .1 ..1. 10 .I ’I 000:.1 ...I . 1 \\ ...01\:0. 01. :: 0.0J.H.I00|.'0:0".~.1'/:::.0 0 1 . . 4 0 0 40 .4 II I A 01 0 0 I : 10 I 0 l. 0.40 0.|.. .... ‘10 .1 o :..I 0014 _§I h . ”0.. 0| 4 .I I 0: 0‘ 00.:04: I :00 4100 Y. 0.000 0:: 4 [0:1 0 0 00 .1. 0 . ... 0 I 0000 00 0 - :0 0: . r 4 0.6.”.0:'.0I.04 -0.-0040””54fit' 0. I . . .. : 00 I .10 I 1 I 4 0 1 u: 0 000.0 040 0 0 ' 0 4 00 .0 I 0 00 '1 000 0 0 I 0 1 1:0: 000I 0 . I. .00 . 00: 0 044“ 0I01l4 400 I I10 00.4 o . . 1 0‘0 40 I I 0 0 . I: 4 00- 00 I 1: v : : 9 I: 0 .0? 0. 0 0.- 0. I r. . . 4‘0. 0.. . .. . '1. . . 1 0. :I 0 4 I I... 040 o . 0 I 11 I . I: 1 0 1 : :A ‘ 1. 1:: 'I: .I .0. I10 1 4.0::00 0‘ .4440. 0.’IP14440‘0‘4: M 11.. .0020‘4'} .r .I.I:.:. . .1 . ..I 0 ..0 .4 : 1:. “.0-...11-‘ 0.4.10.0: .. . .. 0 01. 1.. 1.. 0 100.0. .. .I :. 00( I . . 1.1! 0' 1 . 1 . . 1 .I 0 0 .1 . I 0 . I 01 I .:.—0. .00‘ 1 1.....1I. 0 1 ... 1. I I a 0 . 0 0: . I 0 ..0 Ifi. I 000. 4 ...11‘. I 0.0 I: I 0 11 .001... - . 1 100 . 1 0 . . to : .0 :01 1400.04 . .4l ::1 :-0.0: I 0‘ : 0 0 I 0' 0000\_ .001. .QIII’ - .‘ . I1 0 : I 1 I I: 1.4 0 0 0'4 : 0 . II : : I. .0 . 0 :0 0 a . .00 : I . : . 0 .1 .. 1 ...I- : I : 0 10 .. 0- 0 00 0. I: 041:. 0 0. .: 0 0: 0 400.0 4. 1 . 0 : .: 010.10 .0444 04 0:0:0 0'0 . 004 : I . 00 0 0. 0 . III: 011:: 1.. . .“n....01..u . .0 00H 00 1 .1 1 000 0 00 40 II 1000.0 0 00. .4 4| :0 o I. 0 10 : I .0000 01“ 0 01.0000 .0: 00 I 0 I. I4 4:0 0 .0 00:: 00' 0 . 0 . 1. I u 1 : .40 40 - 0.1 .I : 00. . 1 I0 .04 4: 44 I ... o I 0 - 14 10.14 .1 4’0 .0 4 4.0 I 0 0. 1 0 I. I. 0- 0 0: .00. I: 4 I I0 00 0 . «:000 1. 040 00‘! 4.1 0: 1.0 I. 0 440000 ‘1 :I I 0.0 0011‘0 4 i 00.0.0 . . . > I 41 4 : 1 0 . 0 O 0 '04 I I I: u . .I 0. 1L4. E06 0m. 1‘4 1: a” II ... G 10 0 40:0 0: I 041 I 0.0 \".I.0 0 I 0144' I .400: 4 .l 0 4: I 0 .l 0 0.. 10‘ 4 I b I 04 4 1 I‘ .1 ..0. 1 4 14 0 040 ‘0‘ ‘ 00 1 1 0 4. .1 I . 0.0:. I : 0 Q 4: 1.1 I 0 l 0' 0:0 I 0 V 1.0 0 001.. _ 1.100.. 1 901 1 .0 01 0:: d0: .0 0 .:.- .I - 1 0 .: 1 I. . 0 f 0 I 1 00 04: . AI 4 '._4: . . 0 9 . I 10 0 .: 1V 1 10. 1. V0 0 0 00 . 1 0 01 .11 . 0 I :I 000 . 0 :.A. 0 : Y I .4 0111.00.13.10 1.: . 1 4 0. 1. I 4 : I0. I I. .I I I 00 0' 1 0 1 I I 0 I I _I .. l 1 O 01-1.! 0. . . 0 .11 . 4: 0 0.: 4 0.004 I : II 0 0 000 1 I I 40 0 0 I : I III 0900 .1 ‘001 101:.0 . .0 0 . 0 .0 I . 0 Y 4 0.0.0040 1 I 0010 0 00 1 I 0 I 4. I41. .. ..0 .1 040‘ O : 0:: 0 I 0 0 .. 1 : 0 0 I101 4.194 I .0 1: I 0 : : 00. : 1. ‘0 : 0 0 10 001 . I I - I : I :0. 1 .1 1 1 1 00'..:.III..01.I.1I: 01 I . 00 .:.00 r I. 1.. . o. . 00 0 : : 0 0 o I 00.0: :. 1 n I Y 0 04 0 :: 1.0.06 : I .1. I 4- 4:0 I -4 I 00 : 1 0. 0 000 0. I -. I . 0 0“ .0 . . 1...0 10- : 11' I 1 I . 0000: 0 0 . I 0 I. 0‘0 : o 0 . ..0 0 1 11.1 . .0 . o 1 10 ‘ 0 o 0.4! I. 0.40 0 0 ..0. O. ..01QIOO 1‘00 .40 0.00 I. I. . . o 00 0. .. 0. 0.1.4. I00 1 I14 0 I I 1 O 0 0 I: 0 L 101 0 0 . I - . : : II — I . 0. 0 ‘ I 0 I I 0 . QIII _ __ I. 0 1 1 1 0 I I 1 .0 I: n I 4:0 0 1 0 0. I Y.‘ : III: 1.0.1 70001141 00 0 ’.1: I I 4. 0. 1. .41 . . I. 0 00- .I: 1 :1 . . a .1 . .1 .1! .1 I . I I. : 0... : 0- 1 100 ... oI‘N‘ . .4 1 0 .0 cl. 0 01 .0 00:1 4 0 I I 0.. 1 1 1 I. 4 4 0 I I 0 0 I I: 500 . . 0 I 1. .:II 1 I. 1.. . . . .0 000 I ..0. I . 0.. 4 .0 .. 0.. I: . . I 14 04 1| . . . 0 II I I . I40r:0 l0 . u . 0 I I 0 . .0 . . 4.. I 0 . . . 0 : I .. I 0: . 0 41L .1400 v .00‘0101 .1... 01100: 0001.00-1 0.....I . .0 .10 . 00 : ...... 00.. I. I: II : .. .0.\: .11... . :0... .. . I :-.00. . 0 1 1 . . : .- . 0 0III fi.00h.400. . 114.: 1 .. I. I _ a I. I. .1. I: :0 0 .011 0 : :0... v I 0:0. : 11.0.00. :-: 1 I 1 00. 1 0. .- .....0 1 100.0 05.. 0. .0 I 11.0 00V.‘ In: 00 I: 0. 0'1 0 0 .1 1 .I 1. 0 01 0 0: 1 0: ’1 I II 00 : o 1 10.0.0 .I :4: 001 . . . I 0 .0 .IL 40: I. :0 - I : 0 0 : : 4 I . : 0000‘ 00.44 ..0— 01‘411110:I:rr‘100..0...040‘1. I .I I ..0! «I 00 . .00.... ...0 0104 II 0 a .II a . 0I0 0 .0.. I: 001 l 0 1 4 .: : .I : : . 00:41:00: 4r ‘ : 0 I. 0.... - 1 00. 00. 00.0.00 . 0. : ( 1.0-I..0.- 00 I 4.0 0000. : : 1 II 1 1. a3. .” 40.. 00 01 I:‘0 0. 14.00071 01. 0.. . .:to . .:I 01 .1 1 .0 .0.! : : 000 0: :.0::\ II. 0.0 I 0. ::0. 0: o: ..0: :- o 0 0:- 0“. 14V .1 L. .. .01. ..0 1 . .I .1 0 ..J- I 0 . I .0000 . 0: . . 0 . - I A 0V0 1117090 00 4.0 1 ..4- I. 1 I 1 . I I1 0. .::II- . 1 0:0 . 0 .0 .0 I I : 0 00: I: 0: . 0 . V .0 0 .0 00.0.: 0. 0 4 0. 1 4. I 00. . 1 . - I: 1 0 ‘1 149.1110» 1.4!... 1.1.4.“. .1 01001 0“! I. . - o I : . 0 I 4 0 ..I :4. 0' . . II I 0000. : 00.4 I I : I: : I. | 0: I0 I : I 00 u 0 I 4 . ' 0 O 1 0 1 - 0. 0 .. C 0 4' 0."...10 410. 0000 1:. 1:100 0 000.011 .4 I . 1 0 1 I I I 4 0 V : 01 : 0 I 0 :0. :01 ...-4|! II :I :1 I I I :. I I: III: . .... . .0 0 . 1. . .I..I . . . .1 .0. .4 . 1. I . .1 0 I 0 .:0 :I I .I . I. ‘1 4,044” ....‘I 11.0. 0.4.0 ..L0...4001010O_..I ...0.. I 11000.0 I440. I. I: I 4 1.0000444 ...44 .00400 - I... I0 000.001.. 0. 0'04. 1 l . . :::I: 0| 0... . . 0 1 .0. . 00. . . . 010 0. : . 0 . . . .. : . I : *0 ."o‘aoH I 0. 11111.0. .000“..1.01I104.l.4.00 .40 .10.. 0001 . .0. 4H1. .01 01 .1 00.4:010‘:. - .1004...I:0..I. ..0 400.011 . . .44.. I .000 .. ...... 00.:0 . . .4'... 0 . :0... :11 I . . 0 .1 I :2 . 0.0 o. - I I 1 . 0 .0 1 1. ... .1. I .. . 004.0 0 . . . 0 .. . I . . 004.101.110.400“: 04 010. '0 ‘l 1 0... .40.014Y‘: .4... - 0444."... 10.10 10:0. .. 00004.0...410 0 0.0 .4 Q .41 .I .010:00: .. .40 40 0 It 0 .I o . . 0 I: . . . . : | tktf‘ 09.1 0?.410 n. 1 . .1 00. ..0.. 1. :0 : : 1 o 0 . : . .1 .004 I I. .n p4 004 C‘O.‘1..41 1-“ wrb’.“11 .:. .... .4101} .040 ..0 .w“140.’00 0‘?.4:00000£4t>44:)l.\0040:04 1 ‘III; .Q1004v1044:000004.4'l. . 1 010. I 40.00 I 00. - 0.10 00 I II, 0 wflI00...0L0.0.01:03'1.0—.0Q .1 . . II : 0 I4. .r10 . : I : . . $10401wkfi61fi1h00efaflo... . ...:Mcu 1.0...01L1160101 111110011 0.01 000.0001. .1.. 4:1. I .1. .:u 1.. .:. . . .. VII-11$. 1...H0u..4...0io*\.r1. . grafitrfikfl .11. Ii::0.1...0...1.1%..’:.”1.: u. .0 “1.1.1:: r1:........ ......“me 181.11....F0ILIX ......c 1000.... 1001.... 5 0.1.1.151. 4.41. giggiifbVQétugg 1451.0 walplbsr.:h.10 gtsvntfllgflt vw-r 1 £’ Yx. fifl‘ ' ~ ."‘,_ 1...; ' r LI‘EZ‘I hi 1%.; - ,‘ 2 o 3 ¢ fi}o-gr~l..n CF“ * ‘ Mk. {than cad.» . (Ingmar)! 35 ~31 21:3 ABSTRACT BEHAVIORAL CONTRAST IN DISCRIMINATION LEARNING WITH AND WITHOUT ERRORS By Thomas Lee Kodera Behavioral contrast reliably occurred in pigeons following errorless discrimination training, contrary to Terrace's (l963) obser- vations. In the main experiment, a 60-second green key-light, associ- ated with a variable-interval 30-second schedule of reinforcement, alternated with a 60-second period of extinction when the key was dark. The amount of behavioral contrast produced was directly influ- enced by such aspects of the discrimination training procedure as: (l) the amount of nondifferential exposure to the positive stimulus prior to institution of the discrimination; and (2) the manner in which the negative stimulus was introduced (whether progressively or abruptly). This occurred independently of the number of errors made by a pigeon during acquisition of the discrimination. In a series of control experiments, substitution of a red key-light for the dark key during extinction resulted in greater behavioral contrast, while an increase to three minutes in the duration of the green key-light; associated with reinforcement attenuated the behavioral contrast effect. These results suggest that the use of differential Thomas Lee Kodera reinfbrcement procedures and those variables which influence the ease with which a discrimination is.formed interact to produce behavioral contrast. REFERENCE Terrace, H. S. Discrimination learning with and without "errors." Journal of the Experimental Analysis of Behavior, l963, g, 1-27. BEHAVIORAL CONTRAST IN DISCRIMINATION LEARNING WITH AND WITHOUT ERRORS By Thomas Lee Kodera A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1974 To my grandfather and his vision of a new world. ii ACKNOWLEDGMENTS A sincere statement of appreciation is in order for the guidance and support afforded me by Dr. Mark Rilling, who supervised my thesis research and two very important years of education. To the other members of my thesis committee, Drs. M. Ray Denny and Robert Raisler, I express my gratitude for the wise counsel and stimulation they provided me. I also wish to acknowledge the invaluable assistance of Jacqueline Dawley, Donald Stehouwer, Marcia Vander Kamp, and Michael Daniel in conducting portions of this research. TABLE OF CONTENTS Page LIST OF TABLES ...................... v LIST OF FIGURES ..................... vi INTRODUCTION ....................... l METHOD .......................... 9 Subjects ..................... 9 Apparatus .................... 9 Design ...................... 9 Procedure .................... l0 Baseline Training Phases ........... l0 Discrimination Training Phases ........ 12 Introduction of S- .............. 12 Phase Termination ............... 13 RESULTS .......................... 16 Responses During S+ ............... 16 Responses During S- ................ ~ 26 Relationships Between 5+ and S- Responding . . . . 34 DISCUSSION ........................ 36 CONTROL EXPERIMENTS .................... 40 Method ...................... 40 Results ..................... 41 Discussion .................... 46 DETERMINANTS OF BEHAVIORAL CONTRAST ........... 48 LIST OF REFERENCES .................... 56 iv LIST OF TABLES Table Page 1. Training Procedures for Experimental Groups ..... ll 2. Summany of the Five-Day Progressive Procedure for Introducing S-.. ................ l4 3. Summary of Analysis of Variance of Behavioral Contrast ..................... 25 4. Summary of Mann-Whitney U Tests of Errors (Responses During 5-) ............... 33 Figure 10. LIST OF FIGURES Rate of Responding to 5+ for Individual Pigeons of the Early-Progressive (EP) Training Group for the First and Last Five Days of Each Training Phase ................... Rate of Responding to 5+ for Individual Pigeons of the Early-Constant (EC) Training Group for the First and Last Five Days of Each Training Phase ..................... -. . Rate of Responding to 5+ for Individual Pigeons of the Late-Progressive (LP) Training Group for the First and Last Five Days of Each Training Phase ................... Rate of Responding to 5+ for Individual Pigeons of the Late-Constant (LC) Training Group for the First and Last Five Days of Each Training Phase ....................... Mean Rate of Responding to 5+ for the Early- Progressive, Early-Constant, Late-Progressive, and Late-Constant Training Groups ......... Mean Change in Response Rate to 5+ (Behavioral Contrast) at Each Change of Training Con- ditions ...................... Incidence of Responses During 5- (Errors) Over Sessions Within Discrimination Training Phases . . . Relationships Between Responding During 5- (Errors) and the Magnitude of Behavioral Contrast Produced. . Rate of Responding to 8+ for the Red 5- Control Group for the First and Last Five Days of Each Training Phase ................. Rate of Responding to 5+ for the Three-Minute S+ Control Group for the First and Last Five Days of Each Training Phase .......... vi Page 17 18 19 20 23 24 28 31 42 43 Figure Page II. Mean Rate of Responding to 5+ for Individual Pigeons of the Three-Minute S+ Control Group During Successive One-Minute Periods of the Three-Minute S+ Component ............ 45 vii INTRODUCTION Stimulus control of responding typically derives from situa- tions in which an organism encounters stimuli which signal differing probabilities of reinforcement for emitting a particular response. The consequences associated with responding to each stimulus produce differences in the behaviors manifested in the presence of these stimuli. Such behavioral differences define the development of a dis- crimination between stimuli, whether they are described in terms of differences in response rate, in response latency, or in other measures of response strength. Reynolds (l963, p. 139) has cautioned, however, that "it is an error to conceive of the formation of a discrimination as composed of simple, independent changes in the tendency to respond to one stimulus when the consequences for responding to that stimulus change. Rather, a change in responding and its consequences during the presentation of one stimulus usually brings about changes in the responding to the other stimuli which the organism successively encounters." Changes such as Reynolds described are observed in the phenomenon of behavioral contrast. I). Behavioral contrast is often described within the context of successive discrimination training. For one stimulus, S+, the conse- quences of responding remain constant throughout the experiment: i.e., the schedule of reinforcement associated with that stimulus never changes. However, when S+ alternates with an S- which is associated with nonreinforcement (or reinforcement on a poorer schedule) for the same response, an increase in responding to 8+ occurs relative to the baseline rate of responding during 5+. This response rate increase is the defining characteristic of behavioral contrast. Consistent with Reynolds' characterization, marked changes in responding to 5+ are pro- duced, even though no objective change in the reinfbrcement schedule associated with this component has been effected. Stemming from his early investigations of the acquisition of a discrimination (Terrace, l963a), Terrace formulated an influential theory of discrimination learning to account for behavioral contrast and other indirect effects of discrimination training which he labeled the "by-products of discrimination learning" (cf, Terrace, l972). The basis of Terrace's theoretical position is best gained by reviewing the early empirical evidence from which it is derived. Terrace's (1963a) research questioned the traditional assump- tion that a reduction during 5- of the response which is reinforced during 5+ is a necessary prerequisite to the formation of a discrimin- ation (Keller and Schoenfeld, l950). To invalidate this assumption, Terrace concentrated on those factors which influence responding during 5-, and in so doing noted evidence which systematically related occurrences of the previously reinforced response daring 8-, often designated as errors, to the relative difficulty of the discrimination (e.g., Spiker, l956; Hansen, 1959). Here difficulty was scaled accord- ing to the degree of physical similarity between 5+ and S-. Conse- quently, Terrace (1963a) considered those features of discrimination training which make acquisition relatively easy or more difficult as basic to the specification of those variables which influence the occurrence of responding to S-. The effects of two procedural vari- ables were investigated: the time in the pigeon's experimental history at which discrimination training was instituted, and the manner in which 5- was initially introduced. The variable, time of introduction of discrimination training, influenced the amount of nondifferential exposure to 5+ for each pigeon received prior to the introduction of S-. For some pigeons, discrimina- tion training followed a period of nondifferential baseline training on 5+, with time of introduction thus described as Late. For other pigeons, discrimination training commenced with the first day of train- ing, long before responding to 5+ approached an asymptotic level. This latter level of the variable was labeled Early. The introduction of S- was accomplished either abruptly or gradually across trials. The abrupt introduction of S-, termed the Constant procedure, involved presenting S- at full duration and inten- sity from the outset of discrimination training, with S- differing from S+ only with respect to the relevant stimulus dimension--wavelength or color. The gradual introduction of S-, termed the ProgressiVe proce- dure, involved first presenting S- at reduced intensity for very brief exposures, and then increasing intensity and exposure time with each successive presentation of S-. Crossing the Early vs Late time of initiation of discrimination training conditions with the ProgresSive vs Constant manner of 8- intro- duction identifies the fbur training procedures Terrace (1963a) used to investigate the acquisition of a discrimination: Early-Progressive; Early-Constant; Late—Progressive; and Late-Constant. While each procedure ultimately resulted in discriminated res- ponding, some very interesting differences in acquisition were noted. Foremost for Terrace's immediate purpose was the demonstration that a reduction during 8- of the response that was reinforced during 5+ was not necessary to the formation of a discrimination. The four training groups differed considerably in terms of the number of responses emitted during 5-, with the subjects of the Early-ProgressiVe group displaying errorless or nearly errorless performance. All groups acquired the discrimination nonetheless. Terrace subsequently attributed special significance to the distinction between discrimination learning accom- plished with and without errors based on observed behavioral differences between his "errorless" group, which received Early-Progressive training, and his other groups. 5 When the discrimination was acquired through an Early-Progres- sive procedure, Terrace (l963a) observed that behavioral contrast did‘ not occur. In addition, he asserts that none of the other by-products of discrimination learning occurs either (Terrace, 1972). However, such phenomena are presumed to be ubiquitous for those discrimination training procedures which involve a reduction in responding during 5- (errors). Terrace explains the appearance of behavioral contrast and the other by-products as due to the acquired inhibitory properties of the negative stimulus which ultimately result when there is a reduction in the rate of responding during S- (Terrace, 1966a, 1972). The exper- ience of nonreinforcement for responding during 5- forces the organism to actively withhold responding--an aversive event which energizes responding when 5+ is again presented. Because, by Terrace's character- ization, the errorless learning procedure simply involves passive non- responding, behavioral contrast should not result. A central assumption in Terrace's theoretical position is simply stated: "None of them [the by-products of discrimination learn- ing7 occurs following discrimination learning without errors" (Terrace, 1972, p.251). It is reiterated throughout his papers relating to this subject (Terrace, 1963a, 1963b, 1965, 1966a, 1966b, 1968, 1972), and has been echoed in a current learning textbook (Nevin, 1973) and in a recent major theoretical article (Rachlin, 1973). Detractors from Terrace's position concentrate upon the validity of this assumption. Some research indicates that other by-products of discrimination learning such as emotional reactions and aggression during 5- (Rilling and Caplan, 1973), and attempts to escape from S- (Rilling, Kramer, and Richards, 1973) bear little relationship to the occurrence or nonoccurrence of errors. A similar conclusion is suggested concerning the conditions which produce behavioral contrast. Reynolds (1961), Taus and Hearst (1970), Vieth and Rilling (1972), and Sadowsky (1973) investigated the effects of blackouts on behavioral contrast and found that nonreinforced responding during 5- displayed a rather insignificant relationship with the occurrence of behavioral contrast. The rate of responding during S+ increased even though few responses were made during blackout periods. In addition, the results from Taus and Hearst (1970) indicated that the magnitude of the contrast effect was directly related to the duration of the preceding blackout period. This led them to speculate that one variable important to the production of behavioral contrast is not whether or not the discrimination was learned with errors, but is instead a procedural variable: the duration of 5-. A similar con- clusion was reached by Wilton and Clements (1971) using a horizontal line correlated with extinction as 5-. Consider an alternative interpretation of Terrace's (l963a). observation that behavioral contrast was found only in subjects which learned the discrimination with errors. The crucial observations of Terrace (l963a) were that the Early-Progressive group exhibited vir- tually no errors during the course of discrimination training, and that this same group also failed to show behavioral contrast. Building upon the logic underlying Taus and Hearst (1970), the present inter- pretation is intimately related to the discrimination training proce- dure. Because Terrace's errorless group was also the sole group exper- iencing Early-Progressive discrimination training, the alternative notion must be entertained that it is the Early-Progressive procedure which mitigates against the occurrence of errors and of behavioral con- trast alike. From such a standpoint, it would be fruitful to concen- trate on the procedural differences between the groups and to ask how the variables contained within them influence behavioral contrast and the production of errors. This approach presupposes the possibility ' that the magnitude of behavioral contrast exhibited may increase or decrease in direct relation to the relative effects of each aspect of the training procedure-~that is, behavioral contrast is not an all or none phenomenon. With this in mind, a systematic replication of Terrace's (1963a) study (Experiment I) was attempted. Attention was focused upon the effects of the time of initiation of discrimination training and of the manner of S- introduction, as operationalized by Terrace (l963a), on the magnitude of behavioral contrast produced and on the amount of responding during 5-. Several important modifications of Terrace's procedure were made in the main study to maximize the probability of observing behavioral contrast. First, the stimulus components were shortened in duration from three minutes to one minute each, thus magnifying the potential for sequential effects like transient (local) contrast (Nevin and Shettleworth, 1966; Staddon, 1969; Malone and Staddon, 1973), which are determined by the immediately preceding stim- ulus. Second, following a suggestion by Hearst (1971), a strict per- formance stability criterion was imposed which established reliable baselines against which overall contrast effects were reflected. In order to partition the presumed effects due to training pro- cedure from those due to responding or not responding to 5-, it was necessary to establish more than one errorless group. The key to the design of this aspect of the experiment was evidence that a pigeon's key-peck response rarely occurs when the key is dark (Reynolds, 1961; Friedman and Guttman, 1965; Brown and Jenkins, 1968; Terrace, 1966a; Taus and Hearst, 1970; Nevin, 1973). Since the likelihood of signifi- cant responding to a dark key is low, the probability of producing another errorless group by Terrace's suggested criterion of 25 or fewer responses throughout the course of the experiment (Terrace, 1972) was greatly increased. At the conclusion of the main study, two control experiments were conducted to assess the significance of the major departures from Terrace's original procedure. In the first such experiment, a red key was substituted for the dark key during S-. In the second experiment, the duration of the S+ component was increased from 60 seconds to three minutes, although a dark key was again used during S-. The orientation of this research program was clearly contrary to Terrace's hypothesis that a reduction in responding during 5-, resulting in an active withholding of responding, is essential for the occurrence of behavioral contrast. Instead, the following hypothesis was advanced: the magnitude of behavioral contrast depends upon the procedure by which the discrimination was acquired. Specifically, the magnitude of behavioral contrast exhibited in a discrimination learning situation is directly related to the amount of nondifferential exposure to 5+ received prior to the introduction of S- and to the manner in which S- was initially introduced--whether Progressively or Constantly. By the same logic, the discrimination training procedure directly influ- ences the amount of responding during 5-. This experiment attempted to demonstrate the nature of these presumed relationships. METHOD Subjects Thirty-two adult female White Carneaux pigeons with no prior ex- perimental history served in this experiment. They were maintained at 80 t 2 percent of ad_libitum body weight and individually housed under conditions of constant illumination and free access to water and grit. Apparatus Two standard three-key operant chambers (LVE model 1519) equipped with LVE model 1348 0L stimulus projectors constituted the experimental apparatus. Only the right key, 5 cm. to the right of center and 25 cm. above the floor, operated; the other keys were covered. A minimum effective force of 0.2 N. was required to operate the response key, which was lighted either red or green or remained dark. The key-peck response occasionally resulted in the presentation, according to schedule, or a hopper of mixed grain as a reinforcer. ~ Except during programmed blackouts, the chambers were each illuminated by 2% watt houselights. Extraneous sounds were partially masked by ventilating fans in each chamber. Electromechanical programming and recording equipment was contained in an adjacent room. Design The variables, time of introduction of 5- (Early and Late) and manner of introduction of 5- (Progressive and Constant), were mani- pulated in a 2 x 2 factorial design. Eight pigeons served in each of 10 the four treatment groups. Within-subjects comparisons were made possible by multiple alternations between baseline and discrimination training conditions. Procedure Because only those factors relating to the introduction of S- differed across treatment groups, the conditions for baseline and dis- crimination training were identical for all birds. Consequently, the procedures for each training phase are described below without regard for the specific treatment group designation of each pigeon. The order of these training phases, however, reflects the Early vs Late group distinctions, with the Early groups beginning immediately with dis- crimination training while the Late groups first established baseline I levels of responding before discrimination training was instituted. . The Progressive vs Constant group distinction determined the manner by which exposure to S- was initiated during the first discrimination phase. The basic procedure followed for each group is outlined in Table 1. Baseline training phaseS.--During all baseline sessions, only one stimulus component was available. The response key was green (5+) for 60 second periods, during which time a variable-interval (VI) 30- second schedule of reinforcement was in effect. Each three-second reinforcement presentation was programmed according to the formula described by Catania and Reynolds (1968) for deriving constant prob- ability VI schedules. During reinforcement, the key-light changed 11 TABLE I TRAINING PROCEDURES FOR EXPERIMENTAL GROUPS Experimental Group Early- Early- Late- Late- Phase Progressive Constant Progressive Constant Progressive Constant introduction introduction I of S- of S- BASELINE BASELINE ------------------------ Training Training DISCRIMINATION DISCRIMINATION Training Training Progressive Constant introduction introduction II BASELINE BASELINE of S- of S- Training Training ------------------------ DISCRIMINATION DISCRIMINATION Training Training DISCRIMINATION DISCRIMINATION BASELINE BASELINE III Training Training Training Training BASELINE BASELINE DISCRIMINATION DISCRIMINATION IV Training Training Training Training DISCRIMINATION DISCRIMINATION BASELINE BASELINE V Training Training Training _Training 12 from green to red as the magazine operated. Successive presentations of 5+ were separated by three-second blackouts, during which the key- and house-lights were turned off. Daily baseline sessions terminated fbllowing 25 presentations of S+. Discrimination training phases.--The principal distinction between baseline and discrimination training phases consisted in the interpolation of a stimulus correlated with extinction (5-) between successive presentations of 3+ during discrimination training. Sixty- second presentations of a dark key (5-) alternated with equally long presentations of the green key S+. separated by three-second blackouts, as during the baseline phase. The order of stimulus presentations was determined by a pseudo-random schedule which restricted to two the number of consecutive repetitions of each stimulus. A multiple schedule of reinforcement was in effect during the discrimination training phases, with 5+ associated with a VI 30-second schedule (as it was during baseline training) and 5- associated with extinction. Responses during 5- had no programmed consequences. Daily discrimination training sessions terminated following 25 presen- tations each of 5+ and 5— components, making a total of 50 stimulus presentations. Introduction of S-.--Immediately after the key-peck response was shaped and 25 consecutive responses were reinforced, the training conditions appropriate to each group were instituted. Both Early groups (Early-Progressive and Early-Constant) were placed in the 13 discrimination phase with the next experimental session. The Late groups (Late-Progressive and Late-Constant) established baseline levels of responding to 5+ before the discrimination phase was initiated. The Progressive procedure for introducing S- preceded the first discrimination phase for the appropriately designated groups (Early- Progressive and Late-ProgresSive) according to the procedure outlined in Table 2. All discrimination phase contingencies Were in effect throughout the Progressive introduction of S-, but the dark key was gradually faded in by increasing the duration of its exposure. No special procedure was necessary for the Constant groups since they were exposed to S- at full duration from the first session of discrim- ination training. Phase termination-~The decision to terminate each baseline and discrimination training phase was based upon a comparison of each pigeon's behavior with a stability criterion: a new phase could be initiated after at least ten days of exposure to a given phase, but then only if the standard deviation of the rate of key-pecking for five consecutive experimental sessions was less than two responses per min- ute. Any bird which failed to stabilize within 35 sessions on any phase was dropped from the study. Only one pigeon (from the Early- Constant group) failed to exhibit stable performance by the 35th day of the first discrimination training phase and consequently was replaced. The strict nature of this criterion served a two-fold purpose. . First, while no less arbitrary than most time-dependent criteria, the criterion for establishing stability imposed here added an objective, behaviorally-sensitive, estimate of asymptotic performance to the 14 .xmx xcoc a no: um .mgaocm cmguo Fpo Lou .asocu pocpcoo -m com mg» Lo» apco vosgowcmg ago: mcowumpzapcme xuwmcmucv 1mm .mocmum_moc msgo cw zuwmcmu:_ um .mucoumm cw umpgoamc mp covumgzu um "muoz ¢ om e Ne v NN 3 LT mN om oe _N 5N Nm Nm 8N m NN em m N NN w + 8m $ 8.8 a» 02 A .w .N e Nm P m. w » % N_ me on N, owON N_ Ne NN m_ A NF co, an ace 8N ooN ¢_ QONP m_ on s «N e mp _, NF mm NN N, w m_ om ON F. “M N. NN N_ o_ P. NN op N H o_ pmN _N com ¢_ cmm N gee. m w_ s NF » , m _ N m_ o, m m m _ N , m w a w H N a o e N o , e m N com m coo N coop NV QOON N ooom N F >NHmzmezH onNNHWZNNZH onNNHWZNNZN onNNHmZNNzH onNNHmzuezH onN~mmmmwomm >~m mzh mo >m‘r.0n: 7O ‘r‘v‘~l (0) ' - ((u) 1 50 7/ TwiJJJJH_LLLJJflLJJJuL.LLLLAMLLLJJ__LLLJJ”LLLJJL.LLLLAMLJJJJJ SESSIONS Figure 3.+-Rate of Responding to 5+ for Individual Pigeons of the Late-Progressive (LP) Training Group for the First and Last Five Days of Each Training Phase. 20 LATE CONSTANT GROUP S+ ONLY 9* IS" I S'I' ONLY S+ IS- 2 8+ ONLY " LC 1 ..0-0’ .. TO: uuf“un‘ .g. “5" ‘ 5° :1)... m w «01 W M: YOIWIII-llLuyu¢LL4LuumuLuL WALUHHUHLMLHJLH : Lc 2 \/ 1 5o. ... 0090‘ 1 30: Ill“ ' ' V. (o) " ' ‘1 ”5:1;AALUWUALLJALU#UJLLMAALLQJ4LUWUALH' I30 I- W q no: W 3 'V J». : 90: I45" W M (0 I x I "P M: 901327111 llllJfllllll Mflllll 1111;..11111 LIIH'M! W q 4 ,A . 70 LC N w : so as“ ; mullll “I“. .111121 111144 E L ‘ 90 $1...»- 5 I RATE OF RESPONDING TO 3+ (RESPONSES PER MINUTE) 5 1o: I 503....‘ “I.“ x .. 30:: us) (2) ‘~"“; lalllll lLllllnlllll ‘llllldll‘l lllllvjfiulll lllllndllll '50:: CG II 'I v' : B0:L : IIO: ”w : 90- A... ...-00‘: 703*“ (5251 w (55) - l'ulllll IILWAIIIL Illllvn’alllll lllll:'AIlll IILLJ: I HO- Ova” -L07 Raw. IIO: W 90: ’V'5 M MM 70- 04) III solulLljl lIlJLlnlllll Illnllllllll lllllnnlll_Ll lllll..llll ’— if U, 7’ " IBOELCB ”" IIO: 90: Idwr..~”~ “\f‘raua (”Vf"~” 70- [*una Mu) '~ (m) 5011111111 lllngglllll IIIIIMIIIII IIIIIlelll 11111:;1141 SESSIONS Figure 4. --Rate of Responding to 5+ for Individual Pigeons of the Late—Constant (LC) Training Group for the First and Last Five Days of Each Training Phase. 21 It is important to recognize that the conditions associated with 8+ are never physically changed. Any changes in 5+ response rate are instead due to the effect of S- alternating with 5+ during discrim- ination training. An increase in 5+ response rate at transitions from baseline to discrimination training denotes behavioral contrast, as does a decrease in response rate at transitions from discrimination training to baseline conditions. In all subsequent discussions, the magnitude of behavioral contrast observed was measured by comparing the mean 5+ response rate for the final five sessions of one training phase with the mean 5+ response rate for the first five sessions of the subsequent training phase. Behavioral contrast was displayed in all groups, although there were differences in the magnitude of the effect, as well as in the reliability and permanence of the change in 5+ response rate. In the Early-Progressive group (Figure l), five pigeons clearly showed contrast throughout the course of the experiment. 0f the three remain- ing pigeons, one (EPI) showed behavioral contrast at the first phase change, but thereafter the effect dissipated. Contrast was evident to a small degree in another (EP2) early in the experiment, and only erratically in the last (EP8). All subjects of the Early-Constant group (Figure 2) displayed contrast in all phases of the experiment, with the exception of EC] which showed response rate changes at the start of each new training phase, but which were not maintained more than one or two sessions. Each subject of the Late-Progressive group (Figure 3) and of the Late- Constant group (Figure 4) manifested behavioral contrast across all IIII'II‘IIII‘II 1.1IIIII. 22 training phases. In some cases, the magnitude of the effect was notably small--as, for example, in the records of LPl, LP3, LP5, and LC2--but the presence of the effect was nonetheless evident. Although the data are not presented in Figure 3, contrast was evident in all except one pigeon of the Late-Progressive group from the first day of S- fade-in. The amount of contrast exhibited on each succeeding session increased in direct proportion to the total duration of S-. The mean daily rates of responding to 5+ for each group are reproduced in Figure 5. This figure underscores the observation that behavioral contrast occurred in all groups. In addition, it demon- strates that the groups did not appreciably differ in baseline rate of responding to 5+. For each group, response rates for the initial baseline phase averaged between 50 and 57 responses per minute. Thereafter, the baseline of the Early-Progressive, Late-Progressive, and Late-Constant groups shifted upward to a level approximately ten responses per minute higher than the initial baseline, at which point they remained. An interesting order effect can also be seen in Figure 5. To a significant degree (t= 1.99, g: =126, E < .05) more behavioral contrast was seen overall on transitions from baseline to discrimination training than on transitions from discrimination to baseline training. Figure 6 presents the group means and ranges of behavioral contrast expressed in terms of the change in response rate at each phase change and overall. Negative values denote the observation of induction rather than contrast effects. This figure reveals a trend for Late groups to exceed Early groups in mean behavioral contrast produced. 23 s+ | s+|s -I |s+ ONLY | s+Is-2 | 3+ ONLY | s+ls -3| I " ' jr'A'R'LY—PR' oo‘ne‘ sswe on"oup—" _—'" "'l BCI-' .pco- '15J'ooaflo‘ ;; 70L 4" - E; 00". ‘bfid'cfluot 25 6()- - 'a.‘ EARLY CONSTANT GROUP R 9°" ‘--.~ “m; g 70- as. "I a . 23 50;- 7 2'5 LATE PROGRESSIVE GROUP 0 g 90- M d z _ .. g 70 — I f - a) _ 60 .. W 00".. ..0.. d a: 25 50 M 'I g ['II‘II II: A I ”I LAhTHEIIICONéI-I—J'IIESTANT GaEROUL-HIAP IEIII I ”I“ ”I =__{'] g I00 " 3 90 — .‘Il ' W .1 g 80- M a 70 — '\. \\ ....- ‘ 6(>- - N 5CUEQ444LuLLLLyuIILL_uuuuVAJLIIHLLUUMAILLLILLLLygI111 7/ SESSIONS Figure 5.--MEan Rate of Responding to 5+ for the Early-Progres- sive, Early-Constant, Late-Progressive, and Late-Constant Training Groups- 24 s+l s-1 - S+ONLY S+ONLY -* S+IS-2 OVERALL HEAR 50 4O 30 20 IO 0 MEAN cums: IN RESPONSE RATE T0 “(RESP-1....) JL J» I L l ‘/ 1 l l S+IS—2-S+0NLY l I l I O l I EP EC LP L I V U EP é'c LP Lc GROUP GROUP V l T 1 EP EC LP L GROUP EP EC LP LC GROUP Figure 6.--Mean Change in Response Rate to 5+ (Behavioral Contrast) at Each Change of Training Conditions. 25 While time of S- introduction appears to be the more powerful variable, if this variable is held constant, the Constant manner for introducing 5- tends to produce more behavioral contrast than does Progressive S- introduction. An Analysis of Variance conducted to investigate these trends (see Table 3) indicated that time of S- introduction significantly influenced the magnitude of the contrast effect, with Late 5- intro- duction resulting in greater changes in 5+ response rate. Individual comparisons of group means from the first phase change (5+ [5- l-e>S+ ONLY) and overall by Dunn's Multiple Comparison Procedure (Kirk, 1968, p. 79) disclosed no significant differences at the first phase change. Overall, both the Early-Progressive and the Early-Constant groups differed significantly from the Late-Constant group (p_< .05), display- ing less behavioral contrast than did the Late-Constant group. The difference between the means of the Late-Progressive and Late-Constant groups approached significance. The overall difference between Early and Late groups was significant (p;< .05); the difference between Progressive and Constant groups was very nearly so. Responses during S-.--The dependent variable described as errors refers to the number of responses directed toward the response key during the dark key 5- component. Of interest in the present study was the total number of such responses and their distribution as a function of the discrimination training procedure followed. 26 TABLE 3 SUMMARY OF ANALYSIS OF VARIANCE 0F BEHAVIORAL CONTRAST Source gf_ MS F Between Subjects 3l Manner of 5- intro- duction (M) l 525.93 1.78 Time of 5- intro- duction (T) 1 1493.89 5.06* M x T 1 168.80 0.57 Error (between) 28 296.43 --- Within Subjects 64 Phase change (P) 2 l51.58 2.36 P x M 2 60.48 0.94 P X T 2 155.79 2.42 P x M x T 2 7.40 0.12 Error (within) 56 64.34 --- Total 87 f2_< .05 27 A question which might well be raised at this point concerns the treatment of responses which occurred during programmed blackouts. Should they not also be considered as errors, since they were responses directed toward a dark key? Casual observation indicated that such a consideration was unnecessary; a number of observers associated with this study at various times reported independently that responding during blackouts was the result of a spill-over of activity from an immediately preceding S+ component. Such responding ceased quickly after the onset of the blackout. No responding was noted during blackouts which followed 5- components. The distribution of responses during 5- displayed by each sub- ject is presented in Figure 7, where the data for each group are organized into ordered blocks. At the forward edge of each block is the subject which made the fewest number of responses during S- overall in that group. The remaining subjects are ordered according to the total number of errors they emitted throughout the experiment, from the least to the most. Note that the data are plotted across the cumulative percentage of sessions in each of the two discrimination phases. This transformation was necessary to standardize the dis- parate number of training sessions given each pigeon. One can deter- mine the amount of discrimination training each pigeon received by simply counting the data points. This figure clearly indicates the effectiveness of the Early introduction of S- in reducing the number of errors during discrimin- ation training. The influence of Progressive or Constant 5- intro- duction is more subtle and requires finer analysis. Since the groups 28 m m 1“! humans to s- (muons: Figure 7.v-Incidence of Responses During 5- (Errors) Over Sessions Within Discrimination Training Phases. 29 differed only during the first five days of discrimination training with respect to their exposure to 5-, the occurrence of responding to 5- during this five day period was compared with that of the subse- quent five days. There is a problem with such a comparison though: while the total number of exposures to S- was equal for all groups, the total time of such exposure was not. Constant groups experienced 7500 second of 5—; Progressive groups experienced only 21ll seconds of 5- during the same number of sessions. Consequently, two separate' analyses were conducted, transforming the error data in slightly different ways--as either the rate of occurrence of errors or as the absolute number of errors produced. Regardless of the treatment of the error data, the Early vs Late group comparisons by Mann-Whitney U tests revealed that Late groups displayed more responding to S- during the first five days of discrimination training. Individual group comparisons showed signifi- cant differences between the Early-Progressive and Late-Constant groups. The only other significant difference between groups resulted from the analysis of the absolute number of errors during the first five days: the Early-Progressive group emitted fewer responses to S- than did the Early-Constant group. One major discrepancy in results relating to the method of transforming the error data was noted in the analysis of the Progressive vs Constant manner of introducing S-. The Progres- sive groups made significantly fewer errors during the first five days of discrimination training than did the Constant groups. The [gtg_of occurrence of errors did not differ between the groups. The importance 17111711111 30 of this discrepancy will be noted later in the discussion. No analysis of the occurrence of errors during the second five days of discrimin- ation training revealed any significant differences between the groups. The upper panels of Figure 8 represent the total number of errors made by each pigeon during the first discrimination phase (S+|S- l) and throughout the course of the experiment (S+IS- l and S+IS- 2). Because the Early groups experienced three discrimin- ation training phases while the Late groups experienced only two, the first two discrimination phases alone are included for purposes of statistical analysis. References to the number of errors made through- out the experiment thus must be interpreted as referring to those made during S+IS- l and S+|S- 2. The identity of the pigeon associated with each bar can be gained through reference to the individual data contained in Figures l - 4. The data in Figure 8 are ordered accord- ing to the number of errors made during the first discrimination phase (5+IS- 1). Group means are indicated by horizontal arrows. For the first phase of discrimination training (5+ IS-l), the Early-Progressive group averaged 6.5 errors, ranging from O to 24. The Late-Constant group produced the largest number of errors, averaging l9l.8 and ranging from 5 to 525. Falling between these two extremes were the Early-Constant group, with a mean of 15.0 errors and a range of 3 to 38, and the Late—Progressive group with a mean of 76.5 and a range of O to 132. Applying a criterion suggested by Terrace (l972) that subjects which made 25 or fewer responses to S- be considered errorless, there were thus eight errorless subjects in the Early- 31 / / S+IS-londS+IS-2 / S+IS-l Z/’ <9 r—S‘ as? «, Ill 11 J . . 111.. Is SUBJECTS a d”. p b b _ — — — — h — — p. p p p p b b m m m m ...... .... m m o .385. -w 8 $9.88,. 32 33.23... 55.58 448.22% .3: . -26 SUBJECTS b h 200 - h 0 o 3 400 '- p O 0 5 7.2 .5 3&8 -m 8 $283". _ _ 0 :20 +m ... I: +m .2533: 545:8 figizwm Figure 8.——Relationships Between Responding During 5- (Errors) and the Magnitude of Behavioral Contrast Produced. 32 Progressive group, seven in the Early-Constant group, five in the Late-Progressive, and five in the Late-Constant group during the first discrimination phase. Overall (see upper right panels of Figure 8 labeled S+|S- l and S+IS- 2), the Early-Progressive group produced a mean of 9.1 total errors (range: 0 to 24), the Early-Constant group averaged 2l.9 errors (range: 6 to 39), the Late-Progressive group averaged 85.5 errors (range: 3 to 317), and the Late-Constant group averaged 204.4 errors (range: l5 to 580). At the completion of the experiment, all eight birds of the Early-Progressive group were still errorless, as were six of the Early-Constant group, three of the Late-Progressive group, and four of the Late-Constant group. Considerable caution must be exercised, however, when inter- preting differences between group means, since the assumption of homogeneity of variance, necessary for parametric statistical analyses, is invalid (f_ = l028.8, g:_= 4/7, p_< .01). In view of this, group differences wgié assessed by Mann-Whitney U tests, which are summarized in Table 4. In all comparisons, the Early-Progressive group made significantly fewer errors than did the other groups. The Early- Constant, Late-Progressive, and Late-Constant groups did not signifi- cantly differ from each other. The time at which the discrimination was introduced significantly influenced the number of responses to S-: fewer errors were observed when S- was introduced Early. Such differences were evident during the first discrimination phase as well as overall. The Progressive manner of introducing S- also signifi- cantly influenced the total amount of responding to S-, resulting in 33 TABLE 4 SUMMARY OF MANN-WHITNEY U TESTS OF ERRORS (RESPONSES DURING S-) GROUP VALUE COMPARISON OF U 4P Errors during first discrimination phase (S+|S- 1): EP vs EC 14.5 .p_< .037 EP vs LP 14.0 ‘p_< .032 EP vs LC 7.5 .p < .004 EC vs LP 20.5 n. 5 EC vs LC 21.5 n. 5 LP vs LC 27.0 n s EARLY vs LATE 63.5 p_ < .010 PROG. vs CONS. 92.5 n. 5 Total errors (S+|S- l and S+|S- 2): EP vs EC 12.5 p < .022 EP vs LP 12.5 p_ < .022 EP vs LC 6.0 p_ < .002 EC vs LP 23.0 n. 5 EC vs LC 20.5 n. 5 LP vs LC 24.0 n. s EARLY vs LATE 62.0 p < .010 PROG. vs CONS. 83.0 p_<:.050 Note: EP = Early-Progressive; EC = Early-Constant; LP = Late-Progressive; LC = Late-Constant. 34 fewer errors than did the Constant procedure. However, this differ- ence was not observed during the first discrimination training phase (S+|S- l): the total number of errors made during the initial acquisi- tion of the discrimination was not significantly affected by the manner of S- introduction. Relationships between 3+ and S- responding.--Referring back to Figure 8, this figure details the relationship between the amount of responding during S-, or errors, and the magnitude of the change in response rate to S+. or behavioral contrast for each subject. The left panels relate the number of errors during S+|S- l with the amount of behavioral contrast exhibited at the phase change from this first discrimination training phase to baseline. The right panels relate the total number of errors made throughout the experiment with the mean behavioral contrast produced by each pigeon. Subjects within each group are ordered with respect to the total number of errors emitted during S+IS- l. A direct relationship between errors and behavioral contrast within groups would be represented by a series of bars in the lower panels which reflects the symmetry of the error data in the upper panels. Clearly such was not the case with the present data. Pigeons which produced the fewest errors in each group were as likely to show the greatest amount of contrast as were those which prOduced the greatest number of errors. Spearman rank-order correlation coefficients between the number of errors and the amount of behavioral contrast are indicated 35 in Figure 8 for each group. For the relationship to be considered significant (gf_= 6), the value of r_had to exceed .70. The strongest relationship was -.50, hence it must be concluded that none of the groups differed significantly from zero correlation. DISCUSSION In the process of challenging traditional assumptions, Terrace (1963a) demonstrated that the occurrence of errors was not necessary for the formation of a discrimination. The present study extends this challenge a bit further: it appears that the occurrence of errors is not necessary for the production of the by-products of discrimination training either. In addition, the by-product under investigation in this experiment, behavioral contrast, exhibited no systematic rela- tionship with the number of errors produced while pigeons learned a discrimination. This demonstration suggests that one need not posit separate laws to account for errorless learning as a special case of discrimination learning. To illustrate this point, consider the behavior of the Early- Progressive group. All eight pigeons acquired the discrimination with fewer than 25 responses to S- throughout the course of the experiment. Five of the eight displayed considerable increases in the rate of responding to 5+ during the discrimination phase relative to the base- line rate of responding to 5+. That is to say, five errorless subjects showed behavioral contrast consistently during discrimination training. The remaining three pigeons were less consistent in their behavior, but they gave some evidence of behavioral contrast at some point during the experiment. These results are at considerable variance with Terrace's (1963a), since his errorless group failed to exhibit behavioral 36 37 contrast. Because this latter observation was the basis of Terrace's subsequent theory of errorless discrimination learning, this dis- crepancy in results is of considerable interest. The expressed purpose of the present study was to determine whether the antecedents of behavioral contrast are found within the discrimination training procedure. Thus, the failure to replicate the original Terrace observation was not unanticipated. The remarkable result of this experiment was therefore not that behavioral contrast occurs in errorless discrimination learning, but was instead that pro- cedural variables which affect the acquisition of the discrimination have such long-term influences on subsequent discrimination performance. Early initiation of differential training facilitates learning the discrimination quickly and with relatively few errors. Additionally, it appears to reduce the magnitude of behavioral contrast throughout the experiment from the level which would otherwise occur if a Late procedure were applied. Progressive introduction of S- likewise results in fewer errors during discrimination training, as well as a reduction in the amount of behavioral contrast. One surprising result was the number of pigeons in the three discrimination training groups other than Early-Progressive which were also errorless: six in the Early-Constant group, three in the Late- Progressive group, and four in the Late-Constant group at the end of the experiment. This can perhaps be attributed to the use of a dark key as S-. It may be that the difference between a dark key S- and a green key 5+ is so great that the acquisition of the discrimination is 38 a simple task. The fact that the Progressive or Constant manner of introducing S- was a less powerful manipulation is consistent with this argument, since the gradual introduction of 5- would be unnecessary in easy discriminations. Powerful support for the argument that it is the discrimination training procedure and not the occurrence of errors which determines behavioral contrast is found in the data of these errorless subjects. In all of these cases, behavioral contrast was observed. There were, however, differences in the magnitude of behavioral contrast produced. These differences must logically be attributed to the effects of the training procedure. In the absence of any significant relationship between respond- ing to S- and responding to 5+, the following influences of the dis- crimination training procedure on behavioral contrast and on the occur- rence of errors were noted. First, with respect to the phenomenon of behavioral contrast, Early introduction of 5- decreased the magnitude of the response rate change produced. The effects of the manner in which 5- was introduced, whether Progressively or Constantly, were independent of the effects of the time of introduction variable. The resultant influence of a Progressive introduction of S- was a decrease in the amount of contrast which would otherwise be manifested if a Constant procedure were used. 4 Second, with respect to the number of errors emitted to S- during the acquisition of a discrimination, Early and Progressive introduction of S- resulted in fewer errors respectively than did Late and Constant introduction of S-. The rate of responding to S- was 39 significantly greater for the Late groups than it was for the Early groups. In this experiment, the first five days of discrimination training had a substantial influence upon whether or not a subject acquired the discrimination without errors. Although the rate of responding to 5- did not differ between Progressive and Constant groups, the absolute number of errors produced during the first five days was significantly greater for the Constant groups. This indicates that the Progressive introduction procedure effectively reduced the tendency to respond to S- by simply restricting the opportunity for responding early in the pigeon's experimental history when the ten- dency to respond to S- was greatest. Once the crucial first sessions were completed, responding to S- was similar across all training groups. CONTROL EXPERIMENTS Because the discrepancy of the present results from those of Terrace (l963a) is so basic, two control studies were conducted to explain these differences. Consistent with the orientation of the present research, procedural differences were emphasized. The original Terrace procedure included a red key as 5+ and a green key as S-. There is little reason to suspect that this difference in 5+ and S- key color is sufficient to account for the present results. However, the fact remains that Terrace's S- was an illuminated key and that the stimulus components were each 180 seconds in duration, compared with the dark key S- and 60 second components of the present experiment. These were sufficiently compelling differences to warrant systematic analysis. For one control group, a simple substitution was made in the original design: a red key replaced the dark key used as S- in the main experiment. It was clear from Taus and Hearst (1970) and Wilton and Clements (1971), that 5- duration directly influences the magnitude of behavioral contrast produced. Consequently, for the other control group, atten- tion was focused on 5+ duration to determine the influence of this aspect of the procedure on the rate of responding to S+. Method Eight naive adult female White Carneaux pigeons served in the control experiments, four in each group. They were maintained under conditions identical with the main experiment. 40 41 Discrimination training for both control groups followed the basic Early-Progressive procedure. The training procedure followed with the Red S- control group differed from that applied to the Early- Progressive group in the main experiment only in the respect that during 5- the key was red rather than dark. The red 5- was introduced accord- ing to the schedule described in Table 2. The Three-Minute S+ control group was exposed to stimulus components identical to those used in the main experiment in all respects except S+ duration. Total S+ time was held constant at 25 minutes, but each successive presentation lasted 188 seconds. Eight presentations of 5+ alternated with eight 60-second presentations of the dark key as S-. Results The rates of responding to 3+ for the Red S- group and the Three-Minute S+ group are recorded in Figures 9 and 10, respectively. All four pigeons of the Red 5- group showed behavioral contrast at each change of training phase. Three of the four birds were error- less at the end of the first phase of discrimination training (mean errors = 19.25; range = 2 to 60). Only one bird was errorless at the completion of the experiment (mean total errors = 164.25; range = 2 to 445). Behavioral contrast was very slight, if present at all, in the Three-Minute S+ group. All pigeons in this group were errorless throughout the experiment (mean total errors = 0.75; range = O to 2). Neither group showed a significant correlation between the number of errors emitted and the magnitude of behavioral contrast. 42 RED 53- GROUP S+lS-I S+ ONLY S+lS-2 90' M. a- ‘7()-‘y‘F.~.‘. ‘\\/"'J. ': .5; 5K):: , fl) ofi'fivqrqb :: 3016311151) llllllLlllll 11111, (93); ”03' "' .‘v M _. dh‘pnrt-o 70pm (60) J (52) ‘IE91_1_1_1_1_u_1_1_1_1_1,fn I I n 1 1 I 1 11; R3 (2) M ”V” (o) RATE or RESPONDING TO 3+ (RESP-I“ N 4C) 502,,AI111I11|_1_1$£__LLL1_1#L.LLJ_L/ [ w(SJIOUP MEfiAN , CF‘”‘“O-‘ , ‘7() ._._.’,‘, «rt-outdi — LIA—‘J-J—LJ—l—‘J—Ufl-l—I—u-U—l—l—I—l—‘fl-LJ—LJ—LJ SESSIONS Figure 9.--Rate of Responding to 3+ for the Red S— Control Group for the First and Last Five Days of Each Training Phase. 43 3 MINUTE S+ GROUP S+lS-I 8+ ONLY S+|S-2 TOEW .fN IM H- 50% [3!“ (0) w (0); E 19() o~04'~u~o 70 A W A 50 3'42 (11 (01 M—LLLLUA-l—LJ—l-J—U—I—I/ |'(i[§‘r‘r‘h‘h" l (01 90 3'43 (2) N W ;.L.L.I_L.I_._LI_I_LL”.L.I_LJ_I_ :EW 751:”? 70 M w 3M4 (0) #WW _ GROUP MEANJJJJfiééLI—L'LI RATE or RESPONDING TO 3+ ( “ESP/m“ 90 W 71)]; ‘(PNO‘I‘. IF‘~IP1~., o-o-Ourdpj t, _.I_I_l_l_|.”.L_I..L_|_l__LJ_L.I.J./,.I_LJ_LL_j/ SESSIONS E Figure l0.-«Rate of Responding to 5+ for the Three-Minute S+ Control Group for the First and Last Five Days of Each Training Phase. 44 Comparisons were made between the two control groups and the Early-Progressive group of the main study. These calculations are based on comparisons of the mean response rate for the first five days of one training phase with the mean response rate for the final five days of the preceding training phase. In terms of the mean overall behavioral contrast produced, the Three-Minute 5+ group displayed the least contrast, averaging a rate change of 7.8 responses per minute. The greatest amount of behavioral contrast, on the order of 23.2 responses per minute, was produced by the Red 5- group. The data from the Early-Progressive group for a comparable period of training fell midway between the data for the two control groups: the mean behavioral contrast for the Early-Progressive group was 15.3 responses per minute across the first two changes in training phase. The difference between the Red 5- group and the Three-Minute 5+ group was significant (Mann-Whitney U = 0, p_= .014). Each of the control groups differed from the main Early-Progressive group to a degree which nearly approached significance (Early-Progressive vs Red S-: U = 7, p_= .077; Early-Progressive vs Three-Minute 5+: U = 8, p_= .107). An analysis of the data from the Three-Minute 5+ control group for transient or local contrast was also conducted. The results are presented in Figure 11. For each pigeon, the mean rate of responding to 5+ is recorded for successive one-minute periods of the three- minute 5+ component within sessions. Only the first and last five sessions of each training phase are presented. Transient or local contrast is observed when maximal responding occurs during the first one-minute period of 5+. Birds 3M1 and 3M2 displayed such a pattern of 45 3 MINUTE S+ GROUP 5+ls-I 3+ ONLY 5 +TS- 2 90 80 TO 60 50 90 80 TO 60 50 120 110 90 80 90 80 TO 60 50 MEAN RATE OF RESPONDING T0 5+ (Responses per minute) P EL\\\\ I— 3H1 I 4,111111 111 111 111 Vi‘)\ 111 111 111 11LJJI11 1' (first 111 111 111 111 111 \IRIM AM; nu 1n 1n 111 11111111 nu 11L111 111 I if b I— I _ 3112 b 311 r 1bx’P 111 111JJ1 111 1111411111 111 111 111 j' \\\\.\\\\\3 an nu uuu 11111111111111 u: “I I U’ p.111 1n nu 111 111 {I II 3113 I MI“ h 3114 I W1. “-5 -4 -3 -2 -I A\[A\ W; \*«\\ [LI 111 111 111 111 q ..I ‘ IN 111 111 111 111 11111111111111 111 111 I fir \ f I i)” W 12345 J/IPA W -5 -4 -3 -2 -I W l2345 «IMO “VIII; W -5 -4 -3 -2 -I SUCCESSIVE ONE-MIN PERIODS OF S-I- COMPONENT WITHIN SESSIONS Figure ll.--Mean Rate of Responding to 5+ for Individual Pigeons of the Three-Minute 5+ Control Group During Successive One- Minute Periods of the Three-Minute S+ Component. I .1 11 I111. III 11.1 I .IIIII I 46 responding consistently during S+|5- 1. Bird 3M3 showed transient (local) contrast for three of the five sessions presented. 0n the two other sessions of S+|5- l, 3M3 responded most during the second one- minute period, as did 3M4 throughout the final five sessions of the phase. When baseline conditions were instituted during 5+ ONLY, no consistent changes in response distribution across the three periods were observed for at least the first five days of the phase. In gen- eral, however, the overall level of responding at each period did decline, giving evidence of sustained behavioral contrast during S+|S- 1. When the final five days of baseline training were reached, response rates in each of the one-minute periods of 5+ were quite similar. During 5+|5- 2, only bird 3M2 showed consistent transient (local) contrast. Local maxima for birds 3M1 and 3M4 fell during the second one-minute period of 5+. The response rates of bird 3M3 no longer showed sustained behavioral contrast, although there were some transient (local) effects during the final five days of training. No comparisons of the transient contrast exhibited by the Three- Minute 5+ control group with that exhibited by the main Early-Progres— sive group were possible as the method of data collection at that time was inappropriate to such analyses. Discussion From the results of these two control studies, the fact that Terrace (l963a) failed to observe behavioral contrast appears to be due in part to the use of stimulus components of three-minute duration. The effect of the illuminated key 5- was to increase the magnitude of behavioral contrast observed. In addition to the small amount of 47 sustained behavioral contrast observed in the Three-Minute 5+ control group, transient (local) contrast effects were also found. Consistent with Nevin and Shettleworth's (1966) conclusion that the two forms of contrast appear independent, birds which showed sustained behavioral contrast sometimes did not show transient (local) contrast. The con- verse situation also held true. Transient (local) contrast occurred less reliably within the training situation, subject at all times to considerable individual differences. It must be noted that the Progressive procedure for introducing the illuminated S- was less effective than Terrace's: errors occurred to a substantially greater degree, especially during the second dis- crimination training phase where no special 5- introduction procedures were applied. Although the Red 5- subjects acquired the discrimination less accurately than did Terrace's, there was no indication of a systematic relationship between the occurrence of errors and the occur- rence of behavioral contrast. The three Red 5- pigeons which were errorless at the end of the first discrimination phase showed behavioral contrast at the subsequent phase change. In addition, the one Red 5- pigeon which was errorless at the completion of the experiment showed behavioral contrast throughout the experiment. DETERMINANTS OF BEHAVIORAL CONTRAST As the results of the present experiment demonstrate, the following variables are related to the occurrence of behavioral con- trast. (l) The amount of prior nondifferential exposure to 5+ before discrimination is introduced directly influences the amount of behavioral contrast produced. 0f the two levels manipulated, introduction of the discrimination immediately following the acquisition of the key-peck response reduced the level of behavioral contrast throughout the experi- ment. It is likely that sensitivity to this variable is restricted to large differences in the range of prior exposure, as no significant correlation was found within groups between the number of training sessions and the magnitude of behavioral contrast produced. (2) The procedure by which initial introduction of 5- is accom- plished, whether Progressively or Constantly, affects the occurrence of behavioral contrast. If a Progressive procedure is followed, less contrast results throughout discrimination training than if a Constant procedure, where S- is abruptly introduced, is used. (3) The duration of the stimulus components, within a cur- rently ill-defined range of variation, acts directly upon the magnitude of the behavioral contrast effect. In general, the longer the duration of 5- (Taus and Hearst, 1970; Wilton and Clements, 1971) and the shorter the duration of 5+, the greater the amount of behavioral contrast 48 49 manifested. Extension of 5+ exposure to three minutes in the present study virtually eliminated behavioral contrast following the Early- Progressive introduction of 5-. (4) The physical characteristics of the stimuli selected as 5+ and 5- contribute to the production of behavioral contrast. In this experiment, the similarity between the stimuli, in terms of satura- tion and/or wavelength, directly influenced the amount of behavioral contrast which resulted: the more similar the stimuli, the greater the behavioral contrast. A similar observation was repOrted by Hearst (1969). When the post-intradimensional discrimination training gradients were compared with the correspOnding excitatory gradients, behavioral contrast was observed in the region of 5+. The magnitude of the elevation of the post-discrimination gradient in the vicinity of 5+ was directly influenced by the proximity of 5+ and 5- on the line-tilt dimension. Any analysis of procedural influences on a given phenomenon is inherently incomplete and unsatisfying. The question still persists, what causes the phenomenon? Why should the parameters thus revealed act as they do? With respect to behavioral contrast, a number of theoretical formulations exist which purport to account for the mechanisms underlying the phenomenon. Integration of these formula- tions with the parametric evidence should render a more complete and satisfying analysis. Terrace (1966a) suggested that two factors, notably the value of the preceding stimulus and the fact that differential reinforcement 50 procedures are used, interact to produce behavioral contrast. An analysis of such interactions seems necessary to account for the present results. Mechanisms by which the value of one stimulus can influence responding to stimuli which immediately follow it are often discussed within the context of transient or local contrast (Nevin and Shettleworth, 1966; Staddon, 1969; Malone and Staddon, 1973). The results of the present series of experiments suggest that the value of a stimulus and its effects on other stimuli may in part be ascribed to the process of learning not to respond to 5-. Learning not to respond or to inhib- it responding is influenced in a large part by the absence of reinforce- ment during S-. The ease with which an animal learns that no type of response it can make will produce reinforcement may influence the kinds of behavior observed in the presence of 5- and of 5+ as well. For example, if the physical characteristics of 5+ and 5- are very similar, the tendency to respond during 5- is greater due to generalization of excitation from 5+. Errors will frequently occur. Consequently, more trials will be required before inhibition of responding to S- is complete. One important aspect of the acquisition of a discrimination involves the mapping of predictability relationships between or the formation of expectancies about stimuli, responses, and reinforcement. Predictability can be defined as the probability with which a stimulus signals the opportunity for reinforcement. Usually this reinforcement is contingent upon the occurrence of a particular response. Predicta- bility may be the preferred state of affairs for an animal in an 51 appetitive situation, just as it is in aversive situations (Seligman, Maier, and Solomon, 1971). Conditions which destroy predictability relationships or contradict expectancies or ones in which the contin- gencies are unclear or difficult to discover may be aversive. Aversive situations often elicit classes of behaviors des- cribed as escape (e.g., Bolles, 1970), aggression (Azrin, Hutchinson, and Hake, 1966), or displacement (Falk, 1971) when there is environ- mental support for such behavior. However, the typical successive discrimination paradigm provides little opportunity for expressing such behavior. Instead, the animal is forced to suppress such response tendencies. Under such circumstances, the termination of S- and the response inhibition which it controls and the onset of 5+ produces a post-inhibitory rebound or enhancement of excitatory processes con- trolled by 5+ analogous to Pavlovian induction (Williams, 1965; Nevin, 1973). Returning to the results of the present experiment, how can such an interpretation based on the value of 5- relative to 5+ account fOr the influence of these variables? The first such variable, the amount of prior nondifferential exposure to 5+ before 5- is intro- duced as a component in the discrimination, falls easily within this interpretation. When the discrimination is introduced Early, relatively few prior expectancies must be eliminated. The increase in responding to 5+ due to its differential association with reinforcement proceeds rather independently of the decrease in the tendency to respond during S-. In the Late procedure, however, the animal is allowed to estab- lish expectancies under nondifferential conditions which may not be 52 effective or appropriate in the new situation represented by the dis- crimination. This makes the process of learning not to respond during 5- more difficult since the old expectancies must first be broken down and then must be supplanted by the appropriate expect- ancies. The Progressive introduction of 5-, one aspect of the second variable influencing behavioral contrast, simplifies the discovery of the appropriate relationships between the stimuli and the conse- quences for responding to them by making the differences between 5+ and S- clearer from the start of differential training. For example, the small disc of plastic which serves as the focus for the key-peck response in the standard operant chamber is but one aspect of the entire complex of stimuli present in the training situation. A change in the color of the light projected onto this disc could well be very insigni- ficant or unnoticeable modifications of the environment for the pigeon. The Progressive training procedure makes the stimuli physically as dissimilar as possible, directing attention to the fact that they pre- dict different contingencies. Then over time these physical differences are gradually reduced, transferring stimulus control to stimuli successively more similar. The aversiveness of the Constant procedure for introducing 5- is probably a function of a third variable which influences the amount of behavioral contrast, the degree of similarity between 5+ and S-. As was noted before, the closer 5- is to 5+ on all relevant dimen- sions, the greater is the amount of behavioral contrast observed. Lamar—.— '1 I a, 11.1 [I I'll-Ill! it"IIMIIIII '11)..- [III 1.1111111,‘ 53 Since the excitatory and inhibitory tendencies of stimuli tend to generalize when 5+ and 5- are very similar, considerable exposure to differential reinforcement is necessary to establish stimulus control. The effect of the duration of the stimulus components, the fOurth major influence on behavioral contrast described, can also be discussed as influencing the value of 5- or of 5+. When 5- is short, comparisons with the conditions of 5+ are more frequent. This more frequent alternation could serve to accentuate the differences between the stimuli. Faster learning should be less aversive, resulting in less behavioral contrast. In addition, the animal is forced to remain in the aversive condition of nonreinforcement for a shorter period of time. When 5+ is long, behavioral contrast is attenuated. This could be due to the dissipation with time of the effect of the preceding exposure to S-. Clearly, the foregoing discussion is highly speculative. Con- siderable further research is required to establish its validity. One assumption, previously implicit, is that an inhibitory stimulus of itself need not be aversive. Rather, the course of establishing a stimulus as inhibitory gives rise to potentially aversive situations. To this point, attention has been focused on the value of 5- as a cause of behavioral contrast. This approach appears to account for the effects of all major variables described in this experiment. However, the very fact that behavioral contrast arises only in situa- tions where differential reinforcement procedures are used warrants further consideration. Contrast effects due to the influence of the preceding stimulus do not account for the entire phenomenon. Based 54 upon the evidence from the Three-Minute 5+ control group, in addition to the production of local maxima in responding immediately follow- ing the onset of 5+, there is also a general elevation of response rate during all periods of the 5+ component. Some of the sustained contrast effect may be due to the con- sistent elicitation of higher rates of responding in the presence of 5+. According to Denny's elicitation theory of learning (see e.g., Denny, 1971a, 1971b), such a condition is sufficient for learning to occur. Consistent with this account, 5+ would acquire the control of high rates of responding which initially are elicited by the removal of 5- in an environment which does not support such behaviors as escape, aggression, or displacement. A most productive account to further explain the sustained aspect of behavioral contrast is derived from the autoshaping liter- ature. The indication is that discrimination learning involves inter— actions between respondent and operant conditioning mechanisms (Gamzu and Schwartz, 1973; Hemmes, 1973). In situations in which a stimulus differentially predicts the occurrence of reinforcement, one typically finds the emergence and maintenance of responding typical of that class of responses elicited by the reinforcer (Jenkins, 1973; Moore, 1973). Thus, behavioral contrast results during differential training from an added stimulus-reinforcement contingency which elevates responding above the baseline level maintained by the response-reinforce- ment contingency which operates alone in conditions of nondifferential I ,1 II III IIIM'tlIIl II 55 reinforcement. The nature of the stimulus used as the reinforcer places an important constraint upon this effect, however. Behavioral contrast will be observed only when the operant response and the response elicited by the reinforcer are of the same behavioral class. . I 1 ill ‘1' LIST OF REFERENCES LIST OF REFERENCES Azrin, N. H., Hutchinson, R. R., and Hake, D. F. Extinction-induced aggression. Journal of the Experimental Analysis of Behavior, 19669 _9_s 191-204. Bolles, R. C. Species-specific defense reactions and avoidance learn- ing. Psychological Review, 1970, 71, 32-48. Brown, P. L. and Jenkins, H. M. Auto-shaping of the pigeon's key-peck. Journal of the Experimental Analysis of Behavior, 1968, 11, 1-8. Catania, A. C. and Reynolds, G. S. A quantitative analysis of the responding maintained by interval schedules of reinforcement. Journal of the Experimental Analysis of Behavior, 1968, 11, 327-383 (Whole Part 2). Denny, M. R. Relaxation theory and experiments. In F. R. Brush (Ed.), Aggrsive conditioning and learning. New York: Academic Press, 19 1. Pp. 235-295. (a) Denny, M. R. A theory of experimental extinction and its relation to a general theory. In H. H. Kendler and J. T. Spence (Eds.), Essays in neobehaviorism: A memorial volume to Kenneth W. Spence. New York: Appleton-Century-Crofts, 1971. (b). Falk, J. L. The nature and determinants of adjunctive behavior. Physiology and Behavior, l97l,'§, 577-588. Friedman, H. and Guttman, L. Further analysis of the various effects of discrimination training on stimulus generalization gradients. In D. Mostofsky (Ed.), Stimulus eneralization. Stanford, Ca1if.: Stanford University Press, 1965. Pp. 255-267. Gamzu, E. and Schwartz, B. The maintenance of key-pecking by stimulus- contingent and response-independent food presentation. Journal of the Experimental Analysis of Behavior, 1973, 123'65-72. Hearst, E. Excitation, inhibition and discrimination learning. In N. J. Mackintosh and W. K. Honig (Eds.), Fundamental issues in associative learning, Halifax: Dalhousie University Press, 1969. Pp. 1-41. Hearst, E. Contrast and stimulus generalization following prolonged discrimination training. Journal of the Experimental Analysis of Behavior, 1971, 15, 355-363. 56 57 Hemmes, N. S. Behavioral contrast in pigeons depends upon the operant. Journal of Comparative and Physiological Psychology, 1973, 85’ 171-178. Jenkins, H. M. Effects of the stimulus-reinforcer relation on selected and unselected responses. In R. A. Hinde and J. Stevenson- Hinde (Eds.), Constraints on learning, London: Academic Press, 1973. Pp. 189-206. Keller, F. 5. and Schoenfeld, W. N. Principles of psychology. New York: Appleton-Century-Crofts, 1950. Kirk, R. E. Experimental design procedures for the behavioral sciences. Belmont, Calif.: Brooks/001e, 1968. Malone, J. C., Jr. and Staddon, J. E. R. Contrast effects in maintained generalization gradients. Journal of the Experimental Analysis of Behavior, 1973, 12, 167-179. Moore, B. R. The role of directed Pavlovian reactions in simple instru- mental learning in the pigeon. In R. A. Hinde and J. Stevenson- Hinde (Eds.), Constraints on learning. London: Academic Press, 1973. Pp. 159-188. Nevin, J. A. Stimulus control. In J. A. Nevin (Ed.), The stud of behavior. Glenview, 111.: Scott, Foresman, 1973. Pp.1|5-152. Nevin, J. A. and Shettleworth, 5. J. An analysis of contrast effects in multiple schedules. Journal of the Experimental Analysis of Behavior, 1966, 2, 305-315. Rachlin, Howard. Contrast and matching. Psychological Review, 1973, §Q, 217-234. Reynolds, G. 5. An analysis of interactions in a multiple schedule. Journal of the Experimental Analysis of Behavior, 1961, 4, 107-117. Reynolds, G. 5. Some limitations on behavioral contrast and induction during successive discrimination. Journal of the Experimental Analysis of Behavior, 1963, 6, 131-139. Rilling, M. and Caplan, H. H. Extinction-induced aggression during errorless discrimination learning. Journal of the Experimental Analysis of Behavior, 1973, 29, 85-92. Rilling, M., Kramer, T. J., and Richards, R. W. Aversive properties of the negative stimulus during learning with and without errors. Learning and Motivation, 1973, 4, 1-10. 58 Sadowsky, 5. Behavioral contrast with timeout, blackout, or extinction as the negative condition. Journal of the Experimental Analysis of Behavior, 1973, 19, 499-507. . Seligman, M. E. P., Maier, 5. F., and Solomon, R. L. Unpredictable and uncontrollable aversive events. In F. R. Brush (Ed.), Aversive conditioning and learning, New York: Academic Press, 1971} Pp. 347-400. Staddon, J. E. R. Multiple fixed-interval schedules: Transient con- trast and temporal inhibition. Journal of the Experimental Analysis of Behavior, 1969, 1g, 583-590. Taus, 5. E. and Hearst, E. Effects of intertrial (blackout) duration on response rate to a positive stimulus. Psychonomic Science, 1970 g )2, 265-267 0 Terrace, H. S. Discrimination learning with and without "errors." Journal of the Experimental Analysis of Behavior, 1963, p, 1-27. (a) Terrace, H. 5. Errorless transfer of a discrimination across two continua. Journalof the Experimental Analysis of Behavior, 1963, _6_. 223-232. (b) 7 Terrace, H. 5. Wavelength generalization after discrimination learn- ing with and without errors. Science, 1964, 144, 78-80. Terrace, H. S. Stimulus control. In W. K. Honig (Ed.), 0 erant behavior: Areas of research and application. New York: Appleton-Century-Crofts, 1966. (a) Terrace, H. 5. Behavioral contrast and the peak shift: Effects of extended discrimination training. Journal of the Experimental Analysis of Behavior, 1966, 9, 613-617. (b) Terrace, H. S. Discrimination learning, the peak shift, and behavioral contrast. Journal of the Experimental Analysis of Behavior, l968,_ll, 727-741. Terrace, H. 5. By-products of discrimination learning. In G. H. Bower (Ed.), The psychology of learning and motivation: Advances in research and theory, v. 5. New York: Academic Press, 1972. Pp. l95-266. Vieth, H. A. and Rilling, M. E. Comparison of timeout and extinction as determinants of behavioral contrast: An analysis of sequential effects. Psychonomic Science, 1972, 2], 281-282. Williams, 0. R. Classical conditioning and incentive motivation. In W. F. Prokasy (Ed.), Classical conditioning. New York: Appleton-Century-Crofts.71965. 59 Wilton, R. N. and Clements, R. O. Behavioral contrast as a function of the duration of an immediately preceding period of extinction. Journal of the Experimental Analysis of Behavior, 1971, 16, 425-428. MICHIGAN STATE UNIVERSITY LIBRARIES : II IIIIIIIIIIIIIIIIII 31293 3 274 58