LOGAN: LAWS OF LEARNING (Basic Processes) January 2004 SPATIAL LEARNING: Excitation Current Stimulus Interdependent Variables Contents Paradigmatic/Natural/Asocial/Emotionally-Positive/Interdependent-Operation SECS<>1:00. Choice Behavior: Reinforcement Variables SF****akL SECS<>1:10. Choice: Amount of Reinforcement SF****alL SECS<>1:20. Choice: Delay of Reinforcement SF****amL SECS<>1:30. Choice: Varied/Fixed Reinforcement SF****anL SECS<>1:40. Choice: Partial/Fixed Reinforcement SF****aoL SECS<>1:50. Choice: Probability Learning SF****apL SECS<>2:00. Decision Making: Reinforcement Schedule SF****aRd SECS<>2:10. Decision: Varied/Fixed Reinforcement SF****aRV SECS<>2:20. Decision: Amount/Delay of Reinforcement SF****aKr SECS<>2:30. Decision: Reinforcement/Punishment SF****aRP Systematic Analses of Decision-Making SF****aRZ Relevant uncoded root-abstracts Relevant(?) unabstracted/uncoded literature SECS<>1:00. Choice Behavior: Reinforcement Variables SF****akL . Choice = Preference between outcomes differing in 1 feature RTPT W/20-hr H2O dep, pref H2O vs hypotonic saline - - - - - - - Chi,HM'63+ RTPH After trng R1 w/dep --> food and R2 w/out dep --> food, Kur,KH'68+ pref R1 w/w/out dep - - - - - - - - - - - - - - - - - - Kur,KH'68+ RYCH Pref 4-pellets >1 w/=total amt; pref 1-pellet if larger amt Cap,EJ'89e SECS<>1:10. Choice: Amount of Reinforcement SF****alL ATLH Acq/reversal of choice based on diff Amt(Reinf) - - - - - - Den,MR'55+ RTPT Concl pref incr-decr-f(consum acty) re size of drink tube - Iso,JR'63+ RTPx Pref live rat > varied > constant goal objects - - - - - - Jen,B.'63+ RTPH Acq pref large > small reinf, not-f(dep) - - - - - - - - - Bow,GH'64- RTPH Acq pref re larger Amt(Reinf), acq is-f(both incentives) - Cla,KN'64- PUPH Re pref for large reinf, value hyberbolic-f(FR/delay) - - - Gro,CL'86+ SECS<>1:20. Choice: Delay of Reinforcement SF****amL RYx- R faster, prefer 1 vs 5 sec Delay(reinf) re position - - - Log,FA'52- Reverse delays, gradual reversal in speeds/pref Log,FA'52- PUPH W/concur VI, Rate(OpR) decr-f(Delay(Reinf)) - - - - - - - - Chu,SH'65- RYPH Pref short-chaining+longer-delay >long-chaining+short-delay Log,FA'70+ SECS<>1:30. Choice: Varied/Fixed Reinforcement SF****anL RYPH Determine indif re varied 50% vs constant Amt/Delay(Reinf) Log,FA'65- Above fail to fit w/mean amt or mean incentive (K) Log,FA'65- Interp above re diff tr-to-tr incr >decr K w/large/small Log,FA'65- PUPH Pref mixed > constant Delay(Reinf) at mean, incr-f(range) - Cic,RA'76- PUPH Est indiff re P(Delayed reinf) vs constant delay CRF - - - Maz,JE'91- Above pref P(delayed reinf) incr if signal reinf w/delay Maz,JE'91- Interp: Value P(Rf) is-f(total time in S assoc w/P(Rf) Maz,JE'91- PUPH Choice fixed vs variable Delay/Amt(Rf) is-f average Rate(Rf Bru,D.'94e SECS<>1:40. Choice: Partial/Fixed Reinforcement SF****aoL RTLH Slow acq, no reversal w/CRF vs 67% PRF - - - - - - - - - - Dav,JW'63- TRPH Pref 50%-4 vs 100%-2 pellets; pref 100%-12 vs 50%24 pellets Per,CC'65e R Pref large-PRF vs small-CRF decr-f(absolute Amt(Reinf)) - - Maz,JE'88- Pref CRF vs 50% PRF decr-f(Delay(Reinf)) - - - - - - - - - Spe,ML'90e Above pref decr w/cue after choice re reinf/nonreinf Spe,ML'90e SECS<>1:50. Choice: Probability Learning SF****apL RTPH P(R) approx= P(reinf) w/75% vs 25% correct, w/correction - Bru,E.'39- RTOT W/hi dep and PRF in 1 GB, nonreinf-->run back-and-forth - - Fre,E.'50- HTPO W/gamble sit, match 50-50 P(Reinf), 100% choice w/70:30 - - Goo,JJ'55- H - Matching Prob(R) w/Prob(reinf) in 2- but not 3-choice prob Gar,RA'57- RTCH Choice beh is-f(both recency/freq(Reinf)) - - - - - - - - - Ove,JE'57+ RTPH Near 100% choice w/85:15% reinf sched w/noncorrection - - - Par,A.'58+ Shift to less than 30% choice w/shift to 30:70% sched Par,A.'58+ M C Title: 2-choice beh - - - - - - - - - - - - - - - - - - - Wil,WA'59- M C Devel near-100% pref for 75% vs 25% reinf w/correction proc Wil,WA'60- F P P-matching w/guidance (correction), max w/out (Mouthbreeder Beh,ER'61+ RTPH P(choice) = P(reinf) in 2-choice sit - - - - - - - - - - - Hic,RH'61- RTPH W/interspersed nonreinf trials, choice match 67:33% reinf - Rev,SH'61- hVPO W/P(Reinf) = .75 vs .25, P(hi-P R) incr-f(Amt(Reinf)) - - - Sie,S.'62+ RTPH W/75:25% reinf, match w/lo not hi dep, is-f(rearing environ Sol,S.'62- Title: Event runs and P(R) on lrng rate - - - - - - - - - Pat,JL'63- R Title: Incentive/trng proc on P-lrng - - - - - - - - - - - Uhl,CN'63- PxPH Maximize re spatial R w/P(Reinf) - - - - - - - - - - - - - Gra,V.'64e OYPO W/cockroach, appear to match w/70:30 P(Reinf) - - - - - - - Lon,N.'64- MV Title: Visual P-lrng - - - - - - - - - - - - - - - - - - - Wil,WA'64e M Title: P-lrng - - - - - - - - - - - - - - - - - - - - - - Wil,WA'64e OTPU Turtle match choice re P(rf) w/visual, maximize w/spatial - Kir,KL'65+ RTPH W/noncorrection, 100% pref re 75:25 and 60:40 P(reinf) - - Wei,S.'65e Above results fit with "identity operator" model Wei,S.'65e FTPG P-matching w/guidance (correction) (goldfish) - - - - - - - Beh,ER'66+ RYPH Maximize reinf in P-lrng sit w/immature/adult subj - - - - Lon,N.'66+ RTPH Asymptotic P(choice) > P(reinf) - - - - - - - - - - - - - - Mel,RP'66- C Title: Spatial P-lrng - - - - - - - - - - - - - - - - - - Pol,SF'67+ RTPH W/extensive trng, tend to max reinf w/P = .8/.2 or .65/.35 Cal,RC'68- RTP Title: P-lrng w/correction - - - - - - - - - - - - - - - - Rob,J.'69- RNPO W/70-30 P(ESB reinf)--hi but < 100% R to 70% altern - - - - Joh,RN'70- PCPH Obtain matching choice to Rate(Reinf) vs optimizing w/2 Rs Maz,JE'81- OTPH Armadillo max reinf in 70-30 prob lrng sit, match w/50-50 - Mus,AE'84e HCPO W/concur chain, undermatch re Amt(Reinf) - - - - - - - - - Ban,S.'85e Above undermatching incr-f(value of initial link) Ban,S.'85e Title: Choice in P(Reinf) proc - - - - - - - - - - - - - - Spe,ML'90e SECS<>2:00. Decision Making: Reinforcement Schedule SF****aRd Decision: Preference between outcomes differing in >1 features NOTE: Must insure equal experience w/alternatives PLPH Pref fixed>mixed FR w/=mean; rel pref =f(geom mean Rate(Rf) Fan,E.'67- PUPH W/concur chain, pref terminal VI30 > VI90 - - - - - - - - - Fan,E.'69- Re rel Rate(OpR), above pref decr-f(init-link VI) Fan,E.'69- Interp above re decr expected time-to-reinf Fan,E.'69- PUPH W/concur chain, pref FR-FR sequence w/short init FR - - - - Hal,E.'84+ Maint above pref w/incr 2nd FR, decr Amt(Reinf) re pref Hal,E.'84+ D PH Choice lo-P(hi qual rf) vs hi-P(lo qual) is-f(motiv level) Sim,PV'86+ PUPH W/concur chain, pref FI w/CRF > PRF, decr w/signal re reinf Dun,R.'90+ SECS<>2:10. Decision: Varied/Fixed Reinforcement SF****aRV RTP1 W/small Amt(Reinf) pref PRF w/2 units vs CRF w/1 unit reinf Lev,AM'59e W/large Amt(Reinf) no above pref re equal Amt(Rf) per R Lev,AM'59e RUPH Pref smaller predictable > larger unpredictable reward - - Ste,GD'84- HUPO Pref small-hi-P(Reinf) vs large-lo in 2nd link concur chain Rac,H.'87e Pref large-lo-P(Reinf) in 1st link if lo-P(2nd link) Rac,H.'87e SECS<>2:20. Decision: Amount/Delay of Reinforcement SF****aKr Genz: Include w/in-chain delay in analysis RTPH Choice large > small reinf not-f(3 vs 23-hr food dep) - - - Rey,WF'61+ RARH Choice R1=small-immed reinf is-f(Amt/Delay(Reinf) of R2) - Dav,JW'62- Derive equal Amt/Delay reinf, incr delay = log incr amt Dav,JW'62- RYPH Determine indiff re small-short-delay vs large-long-delay - Log,FA'65- From above, est rel incentive (K)=neg-expon-f(Amt(Reinf) Log,FA'65- From above, est rel incentive (J)=power-f(Delay(Reinf) Log,FA'65- Evidence that Delay(Reinf) incl factor of GB detentn Log,FA'65- Net incentive (K-J) fits most empirical indif/pref data Log,FA'65- RYPH Pref small-short delay > large-long w/short chaining delay Log,FA'70+ Pref large-long delay >small-short w/long chaining delay Log,FA'70+ PUPH Pref R1 w/small-immed > R2 w/larger-delayed reinf - - - - - Rac,H.'72+ W/concur chain, pref R3-->R2 vs R4-->above choice Rac,H.'72+ Above pref incr-f(duration BO B4 S for R2 or R1/R2) Rac,H.'72+ Interp above "commitment" effect to gradient Delay(Rf) Rac,H.'72+ Above incl forced tr at R3/R4, not R1/R2 Rac,H.'72+ PUPH 3/10 subj peck green w/large delayed reinf vs later red - - Ain,GW'74- Re above, peck red w/small immed reinf > no-peck = large Ain,GW'74- PUPH Pref hi-FI-large-reinf > lo-FI-small incr-f(absolute FI) - Nav,DJ'76+ Pref large-reinf incr-f(equal FI sched) Nav,DJ'76+ Above results fit models w/delay gradients or time-to-rf Nav,DJ'76+ PUPH Maint pref large-long delay w/gradual decr small delay to 0 Maz,JE'78+ RVPT Pref large-delayed > small-immed reinf incr w/dep - - - - - CSz,JJ'80e Above pref also incr w/past history of dep CSz,JJ'80e PUPH Pref large-long-delay > small-short-delay w/concur chain - Gre,L.'80+ Above decr-f(absolute delays) w/constant 6:1/3:1 ratio Gre,L.'80+ Above pref incr-f(absolute delays) w/constant 3:2 ratio Gre,L.'80+ Note: Above diff in amt was 3:1 ratio in reinf duration Gre,L.'80+ HVPO W/noise-off reinf, pref immed short-off > delayed long-off Sol,JV'80e Reverse above pref if add delay to both alternatives Sol,JV'80e PUPH Pref small-short-delay > large-long-delay w/short delays - Ain,GW'81+ Reverse above pref w/incr absolute delays, diff constant Ain,GW'81+ RTPH Pref small-short>large-long decr w/prior exper w/long FR/FT Eis,R.'82e Pref small-short>large-long inc-f(dep) w/long long-delay Eis,R.'82e PUPH Pref small-short-delay > large-long-delay w/short delays - Gre,L.'82e Reverse above pref w/incr absolute delays, diff constant Gre,L.'82e PUPH Pref large-long-delay > small-short-delay w/concur sched - Sny,M.'83- Above pref decr-f(absolute delays) w/ratio constant Sny,M.'83- PUPH If pref large-delay > small-immed, no R to reverse choice - Log,AW'84+ HUPO Some O pref small-short-delay > large-long-delay video - - Mil,A.'84+ Above pref decr w/equal incr in delay for both amounts Mil,A.'84+ PUPH Choice small-short-delay vs large-long not-f(dep) - - - - - Log,AW'85+ PVPH W/AuShR, pref large-long-delay incr-f(key-S duration) - - - Pol,A.'85e hxxO Init trng w/large-delay reinf incr pref vs small-short - - Eis,R.'86+ Init trng w/large-hi-effort incr pref vs small-lo Eis,R.'86+ Above init trng did not genz delay-->effort, v/versa Eis,R.'86+ H Pref large-delay > small-immed w/discrete tr/concur VI-VI - Log,AW'86e HUPO Some O pref small-immed > large-delayed, esp if adaptive - Nav,DJ'86- HUPO W/concur sched, pref large-delay = small-immed reinf - - - Kin,GR'87+ Above pref large-delay >small-immed w/long changeover delay Kin,GR'87+ PUPH Pref large-long-delay >small-short-delay w/pretrng re reinf Cha,A.'88+ Above pretrng w/VI sched and varied amount/delay Cha,A.'88+ PUPH Pref small-short-delay > large-long-delay w/w/out food dep Log,AW'88e h W/retarded, pref large-delayed > small-immed decr-f(delay) Rag,SP'88e HVPO Impulsive (re test) < nonimpulsive w/DRL sched w/money rf - VBr,MD'88e PUPH Maint pref large-long > small w/decr delay small to .1-sec VHa,F.'88e Some O devel pref small-short incr-f(delay large-longer) VHa,F.'88e PUPH Pref small-short vs large-long delay reinf is-f(reinf qual) Kin,GR'90+ HUPO Pref small-short-delay vs large-long if maximize reinf - - Log,AW'90e RUPH Pref hi > lo sucrose concen decr-f(delay of hi concen) - - Bra,CM'92+ Above decr in pref w/delay greater w/lo maint diet Bra,CM'92+ Pref hi > lo sucrose concen decr-f(delay of both concen) Bra,CM'92+ Above pref obtain w/2-bar sit w/free/force tr Bra,CM'92+ HUPO Pref small-short-delay/large-long-delay is-f(reinf density) Flo,SR'92+ RUPH Pref large-long-delay > small-short incr-f(N/tr/session) - Has,T.'92+ Above pref w/total daily ration received during session Has,T.'92+ REPH Chronic ip cocaine incr pref small-short-delay > large-long Log,AW'92e RUUH Pref small-short-delay > large-long-delay w/constant cond - Tob,H.'93e Attrib above to diff sensitivity to diff Amt/Delay(Rf) Tob,H.'93e PUPH Pref small-short-delay > large-long-delay incr w/ITI reinf Maz,JE'94- PUPH W/concur chain, pref small-short > large-long w/link1 =FR1 Sie,E.'95+ W/concur chain, pref large-long>small-short w/link1=FR31 Sie,E.'95+ In above, O did not switch R near end of link1 ratio Sie,E.'95+ SECS<>2:30. Decision: Reinforcement/Punishment SF****aRP RUPT Maint stable conflict beh w/OpR1=H2O+incr sh, OpR2 decr sh Ber,R.'62- RTPH After trng w/diff Amt(Reinf), pun larger decr approach tend Bru,JL'65e Note: Above Os also refuse to approach small-reinf arm Bru,JL'65e RTPH Inten punish nec to reverse pref incr-f(diff in reinf) - - Log,FA'69- Note: Above diff in reinf based on diff Amt/Delay(Reinf) Log,FA'69- In above: may prefer reinf + punish but run slower Log,FA'69- Re above, nec inten w/partial punish > w/contin punish Log,FA'69- Interp above re neg incentive value of punish (vs pos) Log,FA'69- RUPH W/concur chain, pref long-delay > short delay pun w/=duratn Del,MZ'78- W/concur chain, pref short- > long-duratn pun w/=delay Del,MZ'78- Pref long-long-delay pun > short-short w/short delays Del,MZ'78- Reverse above pref w/long delays, absolute diff equal Del,MZ'78- Above pref short-short incr-f(duration long delayed pun) Del,MZ'78- RxxH FR trng in runway incr pref large-hi effort > small lo - - Eis,R.'89e FR trng incr pref large-shock > small-no shock Eis,R.'89e HUPO Pref large-delay pos reinf (add pts)>neg reinf (subtract) - Tak,M.'95+ Systematic Analses of Decision-Making SF****aRZ XXXH Genz: Pref larger Amt(Reinf) offset by diff in delay (377) Bit,ME'70+ RYPH Concl decision based on interval(Reinf)=(chain+nonchaining) Log,FA'70+ Concl pref base on minimize chaining > nonchaining delay Log,FA'70+ PVPH W/concur chain, lo transitivity in sched pref (A=B,A>C,B>C) Nav,DJ'72+ XXXX Def: Impulsive = choice small-immed-reward > large-delayed Ain,GW'75- Attrib above to rapid decr in reinf value w/incr delay Ain,GW'75- Above implies change in pref to small-immed over time Ain,GW'75- Describe impulse-control (eg contract/early commitment) Ain,GW'75- RUPH W/diff Amt(Reinf), equivalent delay =power-f(delay) - - - - Ito,M.'82+ PLPH W/concur chain, pref is sq-rt-f(init + terminal VI scheds) Fan,E.'83+ PUPH Concl: Exper w/gradual decr delay decr slope delay gradient Log,AW'84e PUPH W/concur VI-VI, choice is-f(indep/mult-f amount/delay reinf Rod,ML'86+ RTPZ Anal prior exper/motiv on choice small-short vs large-long Eis,R.'87+ xUPx W/pigeon/human, model indifference re amt vs delay(reinf) - Rod,ML'88+ HUPO Re pref amount vs delay reinf, effects indep/multiplicative Log,AW'92e PUPH Est sensitivity to diff in Amt(Reinf) > diff in Delay - - - Tob,H.'93e PUPH Concl choice is-f(ratio) vs diff in reinf duration - - - - Gib,J.'94+ Relevant uncoded root-abstracts Relevant(?) unabstracted/uncoded literature R Title: Perception of spatial relations - - - - - - - - - - Tol,EC'36+ Title: Mult-choice decision beh - - - - - - - - - - - - - Gar,RA'58- RM Title: Cue dominance + direction orient - - - - - - - - - Sha,WA'58+ M Title: 2-choice beh in noncontingent sit - - - - - - - - - Wil,WA'59+ Title: Mult-choice beh w/dummy choices - - - - - - - - - - Gar,RA'61- Title: 2-choice beh w/many alternatives - - - - - - - - - Gar,RA'61- R Title: 2-choice beh w/contingent reinf - - - - - - - - - - Cal,RC'66+ R Title: Blocking/overshadowing w/maze cues - - - - - - - - Die,V.'85e H Title: Reinf dep on self-control - - - - - - - - - - - - - Flo,SR'90- R Title: Pretest exposure to CS - - - - - - - - - - - - - - Gis,P.'90e R Title: Spatial guidance of choice beh - - - - - - - - - - Bro,MF'93e Title: Discounting delayed reward - - - - - - - - - - - - Gre,L.'94e R Title: Ordinal timer in time-place task - - - - - - - - - Car,JA'97+