Gradients of Error-Reinforcement in a Serial Perceptual-Motor Task - No.437 - 1957

Sep 21, 2026

Uploaded by Riley

The study examines whether a statistically significant gradient in error response strength around rewarded responses persists after removing guessing-sequence and probability-bias influences using control groups comparable to experimental groups. It addresses the “spread-of-effect” phenomenon originally reported by Thorndike and reviews competing interpretations, including reinforcement generalization versus artifact accounts from response habits and bias. The work emphasizes the need for controlled comparisons across different reward and experimental frameworks.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages24

Reading Time1 h 0 min

AccessFree to download

I. INTRODUCTION

II. METHOD

A. Subjects

No.437 1957 | Gradients of Error-Reinforcement in a
Serial Perceptual-Motor
Task
By
Melvin H.Marx
University of Missouri | | --- | Melvin H。Marx Vol.71 No.8 # chologicalonographs General and AppliedPricc $1.00Edited by Herbert S.Conrad Published by The American Psychological Association,Inc. ## Psychological Monographs:General and Applied Combining the Applied Psychology Monographs and the Archives of Psychologywith the Psychological Monographs HERBERT S.CONRAD,Editor Department of Health,Education,and WelfareOffice of EducationWashington 25,D.C. Consulting Editors DONALD E.BAIERHAROLD E.JONESFRANK A.BEACHDoNALD W.MACKINNONROBERT G.BERNREUTERLORRIN A.RIGGSWILLIAM A.BROWNELLCARL R.RoGERSHAROLD E.BURTTSAUL ROSENZWEIGJERRY W.CARTER,JR.Ross STAGNERCLYDE H.CoOMBSPERCIVAL M.SYMONDSJoHN F.DASHIELLJoSEPH TIFFINEUGENIA HANFMANNLEDYARD R TUCKEREDNA HEIDBREDERJOSEPH ZUBIN ARTHUR C.HOFFMAN,Managing EditorHELEN ORR,Assistant Managing Editor Editorial Staff:FRANCES H.CLARK,BARBARA CUMMINGS,SADIE J.DoYLE,SARAH WoMACK MANUSCRIPTs should be sent to the Editor. Because of lack of space,the Psychological Monographs can print only the originalor advanced contribution of the author.Background and bibliographic materialsmust,in general,be totally excluded or kept to an irreducible minimum.Statisticaltables should be used to present only the most important of the statistical data orevidence. The first page of the manuscript should contain the title of the paper,the author'sname,and his institutional connection (or his city of residence).Acknowledgmentsshould be kept brief,and appear as a footnote on the first page.No table of contentsneed be included.For other directions or suggestions on the preparation of manu-scripts,see:CONRAD,H.S.Preparation of manuscripts for publication as mono-graphs.J.Psychol.,1948,26,447-459. CORRESPONDENCE CONCERNING BUSINESS MATTERS(such as subscriptions and sales,change of address,etc.)should be addressed to the American Psychological Associa-tion,Inc.,1333 Sixteenth St.N.W.,Washington 6,D.C.Address changes must arriveby the 1oth of the month to take effect the following month.Undelivered copiesresulting from address changes will not be replaced;subscribers should notify thepost office that they willguarantee third-class forwarding postage. CoPYRIGHT,1957,BY THE AMERICAN PsYCHOLOGICAL AssoCIATION,INC. ## Gradients of Error-Reinforcement in a Serial Perceptual-Motor Task¹² MELVIN H.MARX University of Missouri effect.More recent investigators havepresented a wide variety of alternativeinterpretations.These range from thesuggestion that the Thorndike effect issimply one special case of a more inclu-sive generalization-of-reinforcement prin-ciple(9)to the conclusion that the ob-tained gradients bear no essential rela-tion to reward but are artifacts of suchconditions as number-guessing habits andprobability bias in the response series ## I.INTRODUCT1ON HE prinary problem of the presentI study was to determine whether astatistically significant gradient of re-sponse strength of errors around re-warded responses would remain afterremoval of the guessing-sequence andprobability-bias factors by control groupsdirectly comparable to the experimentalgroups. Originally reported by Thorndike (16, (3,4,14,15). 17),the"spread-of-effect"phenomenonhas received substantial empirical verification.Considerable disagreement,how-ever,has developed about its theoreticalinterpretation.Thorndike(16)regardedthe gradient of decreasing intertrial repe-tition of errors with increasing distancefrom a rewarded response as strong evi-dence for the direct strengthening byreward of S-R connections.He saw it asan"independent proof"of his law of The present study is an attempt toprovide data on this problem with moreadequate controls than have hithertobeen used.The necessity for such con-trols is pointed up by the various studieswhich have shown that artifactual gradi-e

FAQs

What is the primary question of the study?

Whether an error-response strength gradient around rewarded responses remains after removing guessing-sequence and probability-bias factors using comparable control groups.

Why are control groups considered necessary?

Prior findings show that artifactual gradients can be produced without reward, and gradients from guessing/bias conditions cannot be directly compared to those from typical reward-with-motivation settings.

How does the introduction relate the study to Thorndike’s spread-of-effect?

It notes Thorndike’s interpretation as evidence for direct strengthening of S-R connections and highlights substantial empirical verification alongside theoretical disagreement.

Sep 21, 2026

Uploaded by Riley