Simulación, persistencia y sustentabilidad

The o r ig in a l P earce-H all (1980) th e o ry appeared to say th a t a re d u c tio n in the asymptote f o r the US could lead to sim ultaneous e x c ita to ry and in h ib ito r y c o n d itio n in g . For example, i f L=1 and th is US becomes f u l l y p re d ic te d , then p re se n ta tio n o f a weaker US, such th a t L = 1/2, w i l l lead to e x c ita to ry le a rn in g dV(p) = @(p) S 1/2 L, and in h ib ito r y le a rn in g d V '(p ) = @(p) S [V (s ) -1 /2 L 3 = @(p) S 1/2 L as V (s) = L . Thus th e re is no o v e ra ll in h ib ito r y le a rn in g ! To remedy

th is anomaly, i t has subsequently been assumed by H a ll (personal communication, 1985) th a t no e x c ita to ry c o n d itio n in g can occur i f in h ib ito r y c o n d itio n in g takes pla ce . This m o d ific a tio n has the u n fo rtu n a te consequence th a t a sim ple a p p lic a tio n o f th e P earce-H all th e o ry is unable to account f o r dow nshift u n b lo ckin g : For th e re to be e x c ita to ry c o n d itio n in g , L must be g re a te r than V (s) - Y *(s ) in the unblocking c o n d itio n . This n e ce ssita te s th a t th e s u rp ris in g n e s s ( ie , L - V ( s ) + V ' ( s ) ) is g re a te r in the blocked c o n tro l group (P— >Sh + Sh, then PQ— >Sh + Sh), as L f o r 2 shocks is h ig h e r than f o r one shock.

L e t us regard th e 2 shocks as separate events ( ie , d iffe r e n t L 's r e fe r to the d iffe r e n t shocks). Pearce and H a ll (1980, p540)

suggested th a t in h ib ito r y le a rn in g to the om itte d second shock in d ow n sh ift u n b lo ckin g , is more than o ffs e t by the increased le a rn in g th a t accrues to th e f i r s t shock. For the absence o f th e second shock to be more s u rp ris in g than i t s occurrence re q u ire s V(p— >2nd Sh) > 1/2 L(2nd Sh). This im p lie s th a t V(p— >1st Sh) > 1/2 L (1 s t Sh). These d iffe re n c e s need to be q u ite s u b s ta n tia l f o r th e re to be much h ig h e r a s s o c ia b ility in th e d o w n sh ift c o n d itio n . This leaves l i t t l e scope f o r Q— >1st shock le a rn in g , as the asymptote w i l l soon be reached, le a d in g to in h ib ito z y le a rn in g . A lso , th e re w i l l be a r e la t iv e ly la rg e amount o f In h ib ito r y c o n d itio n in g to the absence o f th e second shock. I f th e asym ptote f o r th e second shock is low (so th a t th e re is l i t t l e in h ib ito r y c o n d itio n in g ), then th is shock's absence can o n ly e ffe c t a sm all increase in a s s o c ia b ility . U n fo rtu n a te ly the algebra r a p id ly becomes complex, so I do n o t know i f th is fo rm u la tio n o f the P earce-H all th e o ry can account f o r dow nshift u n b lo ckin g ; altho ugh I am

.r

convinced th a t th e re would be o n ly a very narrow range o f parameter values which a llo w s th is fo rm u la tio n to w ork, i f i t can work a t a l l . However, i t is q u ite easy to model th e dow n sh ift unb locking phenomenon, i f i t is assumed th a t B=1 f o r the f i r s t shock, and B=0 fo r th e second.

I t is re le v a n t here, to o u tlin e the experim ents undertaken by Kaye and Pearce (1984b). As described in Chapter 4, these authors (Kaye and Pearce, 1984a) have used th e o rie n tin g response to a li g h t as an index o f th e a s s o c ia b ility (o r "processing a ffo rd e d to the s tim u lu s "). In these experim ents, they looked a t the o rie n tin g response to a l i g h t in a b lo c k in g experim ent. A fte r c lic k e r — >food t r i a l s , th e r a ts receive d l i g h t + c lic k e r — >food t r i a l s , and then li g h t o n ly t r i a l s ; w ith a c o n tro l group having tone— >food t r i a l s in th e f i r s t stage. For th e blocked group, th e re was l i t t l e o rie n tin g to the li g h t d u rin g conpound t r i a l s (u n lik e c o n tro ls ), suggesting th a t the lig h t was ig n o re d , as th e events th a t fo llo w e d i t were w e ll p re d icte d by the c lic k e r . However, a competing response account could e x p la in th is r e s u lt, as th is blocked group made more magazine e n trie s d u rin g compound t r i a l s . Kaye and Pearce's second experim ent employed a s e r ia l c o n d itio n in g procedure, w ith th e li g h t coming b e fo re the c lic k e r . T his s u c c e s s fu lly reduced the number o f magazine e n trie s to m inim al le v e ls , and s t i l l demonstrated le s s o rie n tin g to the li g h t in the blocked group. U n fo rtu n a te ly , in the li g h t o n ly te s t f o r b lo c k in g , th e blocked group had n o n -s ig n iflc a n tly (p=0.08) few er magazine e n trie s than th e c o n tro l group: T herefore the b lo c k in g e ffe c t was n o t shown, a ltho ugh i t was e vid e n t in th e f i r s t experim ent. I t is claim ed th a t these two experim ents, taken to g e th e r, provid e good

support f o r the idea th a t th e li g h t has reduced a s s o c ia b ility d u rin g l i g h t — >food t r i a l s i f i t is presented in compound w ith a p re v io u s ly re in fo rc e d c lic k e r . T his idea o b vio u sly f i t s w e ll w ith both the M ackintosh, and P earce-H all th e o rie s . The Wagner model can account f o r the r e s u lt because th e occurrence o f the food in the c o n tro l c o n d itio n is s u rp ris in g . Hence, th e food is a s u rp ris in g p o s t - t r ia l event w hich in te rfe re s w ith le a rn in g th e c o n te x t-lig h t a s s o c ia tio n , and so leads to g re a te r processing o f the li g h t . P e rs o n a lly , I am n o t to o happy about t h is a n a ly s is because i t e xp la in s th e decrease in a s s o c ia b ility o f th e li g h t , and th e b lo c k in g e ffe c t, by two separate mechanisms. The reduced o rie n tin g response to the li g h t is due to lo s s o f processing a ffo rd e d to th e li g h t , whereas th e b lo c k in g e ffe c t is p rim a rily due to a re d u c tio n in processing the occurrence o f fo o d .

N o tw ith sta n d in g th e in a b i lit y o f the Kaye and Pearce (1984b) experim ents to d is tin g u is h between the th e o rie s , th is approach seems ve ry prom ising. For example, w ith d o w n sh ift c o n d itio n in g ( lig h t — >food + fo o d , then l i g h t + c lic k e r — >food, and a c o n tro l group w ith l i g h t — >food + fo o d , and then li g h t + c lic k e r — >food + fo o d ), th e M ackintosh 'v e r b a l' th e o ry would p re d ic t g re a te r a s s o c ia b ility f o r the li g h t in th e c o n tro l group, whereas th e P earce-H all th e o ry p re d ic ts g re a te r li g h t a s s o c ia b ility in the dow nshift c o n d itio n .

Tha Wagner Acjeoupt X2f Dow nahift U nblocking

Again, th e Wagner th e o ry can employ th e idea o f second-order c o n d itio n in g to account f o r d o w n sh ift un b lo ckin g . A fte r P— >Sh + Sh p a irin g s , i f V (p) > L(Sh 1 ), then th e d ir e c t Q— >Sh a s s o c ia tio n must be in h ib ito r y w ith PQ— >Sh p re s e n ta tio n , but an e x c ita to ry Q— >P

a s s o c ia tio n can a lso be formed so long as P is n o t com ple te ly p re d ic te d . In th e d o w n sh ift unblocking group (P— >Sh + Sh, then PQ— >Sh), P is more s tro n g ly associated w ith shock than in the Low C o ntrol group (P— >Sh, then PQ— >Sh). I t is th is d iffe re n c e w hich enables the Q— >P a s s o c ia tio n to account f o r g re a te r suppression to Q in the unblocking group. This leads to th e p re d ic tio n th a t d o w n sh ift u n b lo ckin g w i l l n o t take place (b u t u p s h ift u n b lo ckin g w i l l ) i f steps a re taken to m inim ize th e Q— >P a s s o c ia tio n . I do n o t know i f th is comparison has been made. However, Rescorla and C o lw ill (1983) have in v e s tig a te d th e p o s s ib ilit y o f Q— >P a s s o c ia tio n s producing the d o w n sh ift u n b lo ckin g e ffe c t. The idea they used was th a t agy p re s e n ta tio n o f P in th e absence o f Q ( a fte r compound c o n d itio n in g but be fore te s tin g Q) should tend to e x tin g u is h the Q— >P a s s o c ia tio n . In lin e w ith W agner's th e o ry , they found th a t p re s e n ta tio n o f P on i t s own o r P— >Sh + Sh t r i a l s , between compound c o n d itio n in g and te s t in both groups, abo lished th e unblocking e ffe c t.

A problem w ith t h is o th e rw ise s a tis fa c to ry account is th a t u nb locking has been demonstrated w ith P— >Sh + Sh, then PQ— >Sh as the unb locking group and P— >Sh + Sh, then PQ— >Sh + Sh as the h ig h c o n tro l group (D ickin so n , H a ll and M ackintosh, 1976). Thus th e P— >shock a s s o c ia tio n must be a t le a s t as stro n g in th e c o n tro l group; so i t is necessary to argue th a t th e Q— >P a s s o c ia tio n is weaker i f two shocks fo llo w PQ p re s e n ta tio n than i f o n ly one does. W agner's (1978) account o f p o s t - t r ia l events s ta te s th a t le a rn in g an a s s o c ia tio n is d is ru p te d by a p o s t - t r ia l event to the e x te n t th a t the p o s t - tr ia l event is s u rp ris in g . I f w ith dow n sh ift unb lo ckin g , we consider the Q— >P a s s o c ia tio n is d isru p te d by the shock(s)

p re s e n ta tio n ( ie , th e p o s t - t r ia l e v e n t), then p re s e n ta tio n o f the p re d icte d double shook must be more s u rp ris in g , and hence d is ru p tiv e , to e x p la in the observed d o w n sh ift u n b lo ckin g . As most dem onstrations o f do w n sh ift unb locking use few stage 1 t r i a l s (see R escorla and C o lw ill, 1983), the occurrence o f th e shocks may be o n ly l i t t l e le a m t about in stage 1, and in t h is way, th e double shock in stage 2 could be le s s p re d icte d than the s in g le shock. In lin e w ith t h is , R escorla and C o lw ill (1983, Experiment 4) found unblocking w ith 2 P— >shock t r i a l s , b u t n o t w ith 8 P— >shock t r i a l s . B u t, H olland (1984) used 48 and 64 P— >food t r i a l s in h is two experim ents w hich demonstrated both up and do w n sh ift u n b lo ckin g r e la tiv e to h ig h and low c o n tro l groups.