The role of content and context in pragmatic reasoning
Abstract
The purpose of this paper is to study pragmatic reasoning in conditional inference tasks. Two experiments were performed in which the subjects ought to choose among alternative responses, rating their confidence in their choice. The linguistic form of sentences (“if…then” or “whenever…then”) and the negativity of antecedent vs. consequent were manipulated in the first experiment, with formal content. The second experiment, with thematic content, manipulated the context or scenario of the problem (causal, temporal or promises/threats) and the probability of factual relation between antecedent and consequent of conditional sentence (deterministic, probabilistic, without relation). The results were discussed within the framework of the theoretical approaches for human reasoning based upon mental scenarios
Full text
The Role o Con en and Con ex in P agma ic Reasoning*
Ma ía Dolo es Valiña, Glo ia Seoane, Mon se a Ma ín, José
Fe nández-Rey & Ma ía José Fe aces
Uni e si y o San iago de Compos ela (Spain)
INTRODUCTION
The aim o his s udy is o de e mine he in luence o a linguis ic o m o condi ional
sen ences and he pola i y o mayo p emises on subjec ´s abili y o eason abou
condi ional ules and hei
con idence in he selec ion o co ec esponses.
The condi ional connec i e “i … hen” exp esses one o he elemen a y
ela ion be ween
an eceden and
consequen p oposi ions. Acco ding o he s anda d logic, a condi ional
o he o m “i p … hen q”
is equi alen o “whene e p … hen q”. Bu he “whene e
… hen” colloquial exp essions s esses he
de e minis ic ela ionship be ween p and
q e ms, and he “i … hen” colloquial exp essions s esses ano he p obabilis ic
ela ionship be ween an eceden and consequen .
The objec i e o he expe imen is o es ha condi ional easoning in he abs ac
con en p oblems a ies as a unc ion o he deg ee o ela ionship be ween an eceden
and consequen o condi ional sen ences and hei pola i y.
_____________
* This pape was p esen ed a he FIFTH CONFERENCE OF THE EUROPEAN SOCIETY
FOR COGNITIVE PSYCHOLOGY - ESCOP, celeb a ed in Pa is, Socié é F ançaise de
Psychologie - UNIVERSITÉ DE PARIS-SUD (Sep embe , 12-16, 1992).
In con as he syn ac ic heo y, we conside ed ha pe o mance
shall
gene ally be
be e in he
"whene e ... hen"
han he "i ... hen"
ules. In addi ion, we
also expec ed
unbe e con idence o he
"whene e ... hen"
linguis ic o m.
EXPERIMENT 1
METHOD
Subjec s
Eigh y-eigh subjec s pa icipa ed in his expe imen (50
emales and 38 males). Fo y-
ou subjec s we e alloca ed a
andom o each o he wo expe imen al g oups. All
subjec s we e
s uden s a he college le el (mean age 18 yea s 7 mon hs), a San iago
de Compos ela (Spain).
The subjec s had no p e ious expe ience wi h his ype o asks. They all ac ed as
unpaid olun ee s.
Design
A 2 x 4 x 4 ac o ial design wi h epa ead measu es on he las wo ac o s was
used. The h ee ac o s we e: linguis ic o m o condi ional sen ence ("i … hen”,
“whene e … hen”), ype o condi ional ules: Modus Ponens (MP), Modus Tollens
(MT), A i ma ion o he Consequen (AC) and Denial o he An eceden (NA
) and he
pola i y, wi h he ou possible combina ions o
a i ma i e and nega i e componen s
(AA, AN, NA NN).
The dependen measu es we e: a) co ec answe o
condi ional p oblems and b)
secu i y in he subjec 's
selec ion esponses.
Ma e ials
Fou e sions o he 32 i em condi ional easoning es we e cons uc ed by he
expe imen e s.
These i ems ep esen ed all he possib1e a ia ions o a basic wo-
p emises
condi ional a gumen s acco ding o he ype o ule (MP, MT, AC and DA) and
he pola i y o he componen s o majo p emises (AA, AN, NA, NN).
The p esen a ion o he 32 i ems was andomized and he
easoning p oblems we e
p esen ed in his o de and in he e e se o de
( IA-1 s. A-2; IB-1 s. IB-2).
Task and P ocedu e
Subjec s
we e assigned a andom o one o he wo linguis ic o ms o
he condi ional
sen ences. The e we e 44 subjec s in each
condi ion and hey we e es ed in wo
expe imen al sessions.
Wi hin each o hese wo main g oups we e
gi en he p oblems in a andomized o de
and hal o subjec s we e se he i ems in
e e se o de . All subjec s we e gi en one
e sion o he es bookle s.
Subjec s ecei ed w i en ins uc ions. The subjec 's ask was assume ha he wo p emises
o each condi ional a gumen a e
ue and, on he basis o his knowledge, o selec o a lis o
h ee al e na i es he conclusion ollowed om he p emises.
A second ask was o a e he subjec i e secu i y in he
se selec ion o esponse o
each p oblem. A i e-poin scale was used, whe e 5 mean he maximum sco e and 1 mean
he minimum
secu i y in he subjec 's selec ion esponses.
These ins uc ions we e ead o he subjec s and hey we e asked o eading hem again
o hemsel es. P io o he es
p oblems he e we e wo p ac ice i ems using condi ional
in e ences. Ques ions we e solici ed om he subjec o hey en
su e ha hey
unde s ood he ins uc ions.
Finally, hey we e allowed o wo k a hei own speed.
Resul s
The e was a signi ican main e ec o he ype o ule (F
(3,258) = 70.56; p < .0001)
and a signi ican in e ac i e e ec be ween ype o ule x pola i y
(F (9, 774) = 5,14; p
< .0001 ). The e was no main e ec in ol ing he linguis ic o m o condi ional sen ences;
howe e , he in e ac ion be ween ype o ule x pola i y x linguis ic o m
was signi ican
(F (9, 774) = 2.56; p < .0067).
Fo abs ac condi ional a gumen s, he mean pe cen age o co ec esponses o
MP, MT, AC and NA ules we e 85.57 %, 57.25%, 25.6%, 26.99%, espec i ely.
The e we e no signi ican di e ences o p oblems in ol ing
AC and DA ules, bu
pos hoc compa isons (Tukey's es ) showed ha AC and DA a gumen s we e signi ican ly
(p< .05) mo e di icul han he co esponding MT and MP. On he o he hand, MP
a gumen s we e consis en ly (p< .05) ea lie han MT
a gumen s.
The compa isons (Tukey's es ) o he mean wi hin each ype
o ule indica ed ha
a gumen s wi h AN majo p emises we e
signi ican ly (p < .05) easie o MT ules. On
he o he hand, he e we e signi ican ly less (p< .05) inco ec in e ences DA when he
subjec s easoned wi h condi ional a gumen s ha ing majo p emises wi h wo nega i e
componen s (NN). Fo AC ules, a gumen s in ol ing a majo p emises wi h AA pola i y
we e, in
ac , signi ican ly (p< .05) easie
han hose
in ol ing he denial o a
componen o he majo p emise.
Finally when hey easoned wi h "i ...
hen"
condi ional sen ences and MP ules,
a gumen s in ol ing AA majo p emises
we e signi ícan ly (p< .05) mo e di icul han
hose wi h AN
and NA pola i y. Bu wi h "whene e ... hen" condi ional sen ences
and MP ules, hese a gumen s (AA) we e easie (p< .05)
han hose in ol ing
AN o NA pola i y.
A inal issue in ol es an examina ion o he e ec s o he h ee manipula ed ac o s on
subjec i e con idence in he co ec
subjec 's easoning pe o mance. A 2 x 4 x 4
(linguis ic o m x
ype o ule x pola i y) ac o ial design, wi h epea ed measu es on he las
wo ac o s was used. The e was also signi ican main
e ec s o ype o ule F (3,258) =
72.73; p < .001)
.
Tukey's compa isons (p < .05)
e ealed be e secu i y o he MP
p oblems han hose MT, AC and DA. Mo eo e , he con idence ha a subjec has abou
his logical pe o mance was signi ican ly g ea e (p < .05) o AC and DA p oblems han
hose MT ules.
The e was also a signi ican ype o ule x linguis ic o m in e ac ion (F (3,258) = 3.52; p<.
0156).
The compa isons (Tukey's es ) o he means o “i … hen” a gumen s and
"whene e ... hen"
condi ionals o each ype o ule, showed ha
MP, MT
and AC
ules wi h "i ... hen" sen ences we e signi ican ly (p< .05)
easie han he co esponding
"whene e ... hen"
p emises. The DA ule was signi ican ly (p < .05) easie o
"i ...
hen”
p emises.
EXPERIMENT 2
The aim o his expe imen is o explo e he impo ance o :
a) ype o hema ic con en
and b) ac ual ela ionship be ween he p and q e ms o condi ional sen ence, on he subjec ´s
pe o mance and on he subjec 's co ec esponse con idence.
The expe imen 1 ailed o show any signi ican di e ence be ween he pe o mance on
he “i p hen q” and “whene e p hen q”
condi ional ules. In his expe imen , he ma e ials
we e composed o hema ic elemen s in a cohe en ela ionship be ween an eceden and
consequen o majo p emise. I is possible o hypo hesize ha such cases
should p oduce
acili a ion
pe o mance.
To assess his hypo hesis, in expe imen 2 he deg ee o ac ual
ela ionship be ween he
wo componen s o majo p emises was man ipula ed.
We also assessed he e ec s o di e en hema ic con en . I people eason using syn ac ic
ules nei he ype o con en should be mo
e e ec i e in i sel . In con as , i people eason
using
p agma ic easoning schemas (Cheng & Holyoak, 1985, 1989; Cheng, Holyoak,
Nisbe & Oli e , 1986)
pe o mance on he causal and p omises/ h ea s
p oblems should be
supe io o pe o mance in he empo al
condi ion.
METHOD
Subjec s
The subjec s we e nine y-six s uden s (47 males and 49 e
males) a he college le el
(mean age 18 yea s, 5 mon hs) a
San iago o Compos ela (Spain).
They all pa icipa ed as unpaid olun ee s in his s udy. None
o hese subjec s ha e
ecei ed any ins uc ion in logic and
none o hem had pa icipa ed in he expe imen 1.
Design
A
3 x 4
x 3 ( ype o con en x ype o condi ional ule x p obabili y o ac ual ela ion
be ween an eceden and consequen ) ac o ial design was used. The ype o con en
ac o had h ee le els: Causal con en , Tempo al and P omises/Th ea s con en .
This
ac o was manipula ed be ween
subjec s, wi h 48 pa icipan s andomly assigned o each
expe imen al
g oups.
The epa ead measu es on he las wo ac o s was used. The second
ac o ( ype o
condi ional ule) was simila o he e
xpe imen 1. Finally, he las ac o (deg ee o ac ual
ela ion be ween an eceden and consequen o majo p emise) had
h ee le els: De e minis ic,
P obabilis íc o Wi hou Rela ion.
Ma e ials
Six a ia ions o he 24 condi ional easoning es we e
used. Tes bookle s we e
cons uc ed each con aining an
ins uc ion page and he condi ional easoning p oblems
p esen ed on sepa a e pages.
The e we e 24 condi ional a gumen s o each o h ee di e en con ex s: a)
Causal (IIA-1, IIA-2; g.“I you all o he wa e hen you will ge we you sel ”,
b) Tempo al (IIB-1, IIB-2; g.“i oday is monday hen yes e day was sunday”
and c) P omises/Th ea s (IIC-1, IIC-2; g.“I you ha e he snack hen I ake you
o a walk”).
As in expe imen 1, each bookle was p epa ed con aining condi ional a gumen s
acco ding o he ype o logical ule (MP, MT, AC and DA). Finally, he
condi ional sen ences di e ed acco ding o he p obabili y o ac ual ela ionship
be ween an eceden and consequen o majo p emise: a) De e minis ic ( g. “i
you pu you hand in o i e hen you will bu n you sel ”), b) P obabilis ic ( g. “i
you s udy hen you pass”) and c) Wi hou Rela ion ( g. “i you a e ugly hen you
a e pleasan ”).
These sen ences we e based on p e ious expe imen al wo k. 100 unde g adua e
Psychology s uden s om he Uni e si y o San iago de Compos ela we e equi ed o
w i e 20 condi ional sen ences wi h causal con en , 20 condi ional sen ences wi h
empo al con en and ano he 20 wi h p omises/ h ea s. A u he 100 simila subjec s
ook pa in a pos e io s udy. The subjec ´s ask was o a e he p obabili y o ac ual
ela ion be ween an eceden and consequen o hese condi ional sen ences. Finally,
he condi ionals wi h a maximum deg ee o ag eemen be ween he subjec s we e
selec ed by he expe imen e s o he easoning p oblems es o expe imen 2.
The o de o p esen a ion o he i ems wi hin each bookle was andomized as hose
o expe imen 1.
Task and p ocedu e
In gene al, he p ocedu e was he same as ha used in he p e ious expe imen . The
s udy was un in h ee expe imen al sessions. Thi y- h ee subjec s we e alloca ed a
andom o each o he h ee expe imen al g oups (causal con en , empo al con en o
p omises/ h ea s) in he condi ional es p oblems.
As in expe imen 1, subjec s we e equi ed o selec a
conclusion which necessa ily
ollowed om he p emises o each condi ional a gumen and o a e he secu i y in hei
selec ion o esponse. A i e-poin scale was used.
Subjec s we e gi en unlimi ed ime o ca y ou he ask.
Resul s
An ana1ysis o a iance o he co ec ness o subjec 's answe
o each o he es i em
indica ed ha he e we e signi ican main e ec s o : ype o ule (F (3,275 = 38.28);
p< .0001), he ype o con en (F (2, 186) = 4.92; p<. 0093) and p obabili y o ac ual
ela ionship be ween an eceden and consequen o he majo p emise (F(2, 186) =
9.3; p< .0001).
Fo hema ic condi ional a gumen s, he mean pe cen ages o co ec esponses o MP,
MT, AC and DA ules we e
83.30 %,
60.57 %, 44.97 % and 36.76%, espec i ely. The
compa isons (Tukey´s es )
o each ype o ule showed ha MP condi ional a gumen s
we e signi ican ly (p< .05) easie han he co esponding MT,
AC and DA a gumen s. On
he o he hand, MT condi ional a gumen s we e
consis en ly easie (p< .05) han AC and DA
p oblems and,
inally, AC i ems we e easie (p
< .05) han hose o DA condi ional
p oblems.
Tes o he simple main e ec s showed ha subjec ´s pe o mance wi h causal
con en was signi ican ly (p< .05) be e han he empo al and p omises/ h ea s
con en s. The e we e no signi ican di e ences o p oblems in ol ing empo al
con en o p omises/ h ea s included in he p emises.
Tukey´s es compa ison o he ac ual ela ion be ween an eceden and consequen
o condi ional sen ence (p< .05) demons a ed ha he subjec s pe o med be e when
hey easoned wi h de e minis ic ela ions be ween he wo componen s o he majo
p emise.
The e we e also a signi ican in e ac i e e ec be ween ac ual ela ion x ype o
con en (F (4, 186) = 2.80; p< .0273 and be ween ac ual ela ion x ype o ule (F (6,
558) = 13.79; p< .0001. Tukey´s compa isons (p < .05) indica ed signi ican ly be e
pe o mance on he de e minis ic ela ion han on ei he o he o he wo ypes o
ela ions, wi h causal con en and p omises/ h ea s in he p emises. Bu , wi h empo al
con en , he p obabilis ic ela ion be ween an eceden and consequen showed he
g ea es pe o mance (p < .05).
On he o he hand, inspec ion o pai ed compa isons o he ac ual ela ion x ype o
ule e ealed ha subjec s make mo e co ec esponses wi h he MP, MT, AC ules
in he de e minis ic ela ion (p< .05). Bu also indica ed ha judging conce ning DA
in e ences a e o en mo e co ec wi h p obabilis ic ac ual ela ion be ween he wo
componen s o majo p emise (p< .05).
As in expe imen 1, subjec i e secu i y was examined. As in ea lie expe imen he e
was a signi ican main e ec s o he ype o ule (F (3, 279) = 14.90; p< .0001).
Tukey´s compa isons (p< .05) o he main e ec o he ype o ule (F( 3, 279) =
14.90; p< .0001). Tukey´s compa isons (p< .05) o he main e ec o ype o ule
also showed g ea e secu i y on he MP p oblems han on he o he h ee ules (p<
.05). No o he compa isons we e signi ican .
Las ly, he e we e signi ican main e ec s o p obabili y o ac ual ela ion (F ( 2, 186) =
10.29; p < .0001) and a
signi ican in e ac ion be ween ac ual ela ion x ype o con en (F
(4, 186) = 3.72; p < .0062). Tukey's compa isons (p < .05) o
ac ual ela ion showed
signi ican di e ences in he de e minis ic condi ion and he o he wo ela ionships.
Inspec ion o pai ed compa isons o he ac ual ela ion x ype o
con en in e ac ion
showed ha wi h empo al con en
he subjec s a e mo e he secu i y in he selec ion o
esponse in he p obabilis ic ela ionship condi ion. Ne e heless, he de e minis ic ela ion
showed he highes sco es on each o he wo emaining
con en o p oblems.
REFERENCES
B aine, M.D.S. (1978). On he ela ion be ween he na u al logic
o easoning and s anda d
logic. Psychological Re iew, 85(1), 1-21.
By ne, R.M.J. (1989). E e yday easoning wi h condi ional sequences. The Qua e ly
Jou nal o Expe imen al Psychology, 41A(1),
141-166.
Cheng, P.W., & Holyoak, K.J. (1985). P agma ic easoning
schemas. Cogni i e
Psychology, 17(4), 391-416.