yesterday, today and tomorrow

8
Acta Psycbologica 43 (1979) 469-476 0 North-HollandPublishing Company YESTERDAY,TODAYANDTOMORROW* Benny SHkNON Dept. of F3ychologv,Hebrew Univemity of Jerusalem,Israel Received October 1978 1260 American subjects were timed as they responded to one of the three questions (‘Whatday of the week is today?’ (Tad question), ‘What day of the week was yesterday?’ (Y question), and What day of the week will tomorrow be?’ (Tom question)) at one of three times of the day (early morning, midday and the late evening). Response times as a function of the day on which the question was posed defmed an inverted U-curvefor all questions in all three conditions. The Tod question always triggered the fastest responses. In the morning, responses to the Y question were faster tban responses to the Tom questions, whereas in the evening the converse was the case. At midday responses to the Y question were faster than responses to the Tom question at the beginning of the week and slower than those at its end. The patterns of introspection associated with the tesponacs also defined reversedU-functions. The results are interpreted in the framework of a multichannel spreadingactivation model. In an original study, Koriat and Fischhoff (1974) timed Israeli subject8 as they responded to the question ‘What day of the week is today?‘. They found that difficulty to respond to this question defined an inverted U-function with respect to the day on which the question was posed. Subjects were slower and less accurate in the middle of the week; they were faster and more accurate in its beginning and end. In order to account for the inverted U-function, Koriat and Fischhoff proposed a two stage model. According to the model subjects first * This project coull have never materialized without the help of my assistants. Thomas Turner coIieoted the data. Thomas Heggestad helped in the anslyses and overalladministxa- don. Lea shrtil conducted the tests for quadratic trends. I am grateful to ali of them. Thanks are due also to the Editor of this Journsl and the anonymous referees for their c&Mm and suggestions. Requests for mprlnts should be sent to Benny Shanon, Department of Psychology, Hebrew University Jerusalem,Israel.

Upload: benny-shanon

Post on 25-Aug-2016

214 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Yesterday, today and tomorrow

Acta Psycbologica 43 (1979) 469-476 0 North-Holland Publishing Company

YESTERDAY,TODAYANDTOMORROW*

Benny SHkNON Dept. of F3ychologv, Hebrew Univemity of Jerusalem, Israel

Received October 1978

1260 American subjects were timed as they responded to one of the three questions (‘What day of the week is today?’ (Tad question), ‘What day of the week was yesterday?’ (Y question), and What day of the week will tomorrow be?’ (Tom question)) at one of three times of the day (early morning, midday and the late evening). Response times as a function of the day on which the question was posed defmed an inverted U-curve for all questions in all three conditions. The Tod question always triggered the fastest responses. In the morning, responses to the Y question were faster tban responses to the Tom questions, whereas in the evening the converse was the case. At midday responses to the Y question were faster than responses to the Tom question at the beginning of the week and slower than those at its end. The patterns of introspection associated with the tesponacs also defined reversed U-functions. The results are interpreted in the framework of a multichannel spreading activation model.

In an original study, Koriat and Fischhoff (1974) timed Israeli subject8 as they responded to the question ‘What day of the week is today?‘. They found that difficulty to respond to this question defined an inverted U-function with respect to the day on which the question was posed. Subjects were slower and less accurate in the middle of the week; they were faster and more accurate in its beginning and end. In order to account for the inverted U-function, Koriat and Fischhoff proposed a two stage model. According to the model subjects first

* This project coull have never materialized without the help of my assistants. Thomas Turner coIieoted the data. Thomas Heggestad helped in the anslyses and overall administxa- don. Lea shrtil conducted the tests for quadratic trends. I am grateful to ali of them. Thanks are due also to the Editor of this Journsl and the anonymous referees for their c&Mm and suggestions.

Requests for mprlnts should be sent to Benny Shanon, Department of Psychology, Hebrew University Jerusalem, Israel.

Page 2: Yesterday, today and tomorrow

470 B. Shanon~Yesruday,, today and tomorrow

determine the part of the week corresponding to the target day: beginning, middle or end. Only then do they continue and determine what the specific day is. The curvilinear relationship between difficulty of retrieval and the ordinal position of the target day is due primarily to the first stage. “On Sunday or Friday,’ when memory of the past Sabbath or anticipation of the coming Sabbath are vividly present, the second, more analytic stage of the search may be eliminated entirely” (Koriat and Fischhoff 1974: 205).

Koriat and Fischhoff conducted their investigation among Israeli Jews whose work week consists of six days, with the Sabbath being the only day of rest; the investigations were held on work days only. Further, in the language of this community, Hebrew, the days of the week (except for the Sabbath) do not have names and they are denoted by ordinal number instead (e.g., Sunday is ‘first day ‘). It is interesting, then, to examine whether the patterns found are replicable in a differ- ent social environment, in a linguistic community which does not employ an explicit numerical system, and - further - whether they extend over the entire week, not just work days. Indeed, other extensions are feasible as well: the examination of different temporal queries and the conduction of the investigations in different times during the day; both are introduced in the present study. These extensions will, in turn, provide a basis to test the interpretation that Koriat and Fischhoff offer to their own data.

The elxperiment

Methsd l%e procedure followed that of Koriat and Fischhoff with a few modifications

which are detailed below. Ss were timed as they answered one of the following questions: What day of the week is today?’ (Tod question), ‘What day of the week was yesterday?’ (Y question), and ‘What day of the week will tomorrow be?’ (Tom question). Each of these questions was investigated three times: in the morning (7-9 a.m.), at midday (12 noon-2 p.m.), and in the evening (9-l 1 p.m.).

Time meaeurements were made by means of a stopwatch from the end of the queotion to the production of the answer. Each S was presented with only one exp?rhnenud question preceded by the question ‘What is the capital of Maine?‘, which was poacd in order to accustom Ss to the questioning procedure. Following

’ These particular days apply to the working and rest pattern in Israel. For the American roesetu. One sboukl substitute Monday for Sunday without any effect on the rationale.

Page 3: Yesterday, today and tomorrow

8. Sheno~/Yesterdey. to&y and tomorrow 471

Table 1

Response times for three questions in three times of day (in msec).

Time of dsY

Experiment Day Quadratic trend question Sun. Mon. Tues. Wed. Tburs.Fri. Sat. F p

-

Morning Y. 1.77 2.01 2.19 2.35 2.26 2.15 1.80 20.71 0.001 TOd 1.58 1.60 1.64 2.17 1.89 1.61 1.38 14.80 0.001 Tom 1.73 2.20 2.52 2.89 2.33 1.22 1.90 5.99 0.025 Mean 1.70 1.94 2.12 2.47 2.16 1.66 1.69

Noon Y. 1.78 1.80 2.25 2.44 2.43 2.07 1.72 7.87 0.01 Ted 1.05 1.39 1.61 2.11 1.84 1.12 1.26 11.66 0.001 Tom 1.55 2.10 2.60 2.10 1.88 1.72 1.65 26.20 0.001 Mean 1.46 1.76 2.15 2.22 2.06 1.64 1.54

Evening Y. 2.80 3.07 3.31 3.76 3.74 2.78 2.57 13.50 0.001 Ted 2.08 2.20 2.41 2.73 2.22 2.05 1.74 0.40 0.025 Tom 2.57 2.60 2.93 3.07 2.83 2.53 2.53 3.62 0.06 Mean 2.48 2.62 2.88 2.19 2.93 2.45 2.28

the experimental question the S was asked to freely describe how he had deter- mined the identity of !he day in question. The S’s introspection was then classified by the experimenter as belonging to one of several types, to be noted below.

The design differed from that of Kdriat and Fischhoff in that a controlled number of Ss participated in each condition. 20 different Ss giving correct responses were

0.5

t

---- YESTERDA’I

- TOC-AY

--- TmORROW

d”_L._.. I _..L-_-_L... . L--.-L M T W fh F Sat

DAYS OF THE WEEK

Fig. 1. Response times to Today, Yesterday and Tomorrow questions (Morning).

Page 4: Yesterday, today and tomorrow

472 B. Shanan~Yesteday, to&y and tomottvw

3.0

r

o.i , , yyiy , H T W l-n F SAT

OAYS OF THE WEEK

Fig. 2. Response limes to Today, Yesterday and Tomorrow questions (Noon).

recruited for each part of day, for each time of day, for each question, Le., the data are based on correct answers collected from 1260 (3x20~3~7) Ss. Ss were Harvard and MIT students recruited on campus grounds.

Results

Respansa times The mean response times are Riven in table 1 and figs. 1,2 and 3. On the abscissa

is the day on which the question was posed (which is the same as the target day for the Tod question, but is one day after and one day before the target days of the Y and Tom questions, respectively).

In a four way analysis of variance, the factors Question and Day and their interaction all reached significance beyond the 0.0001 level (F(2, 1197) = 85.72, F(6,1197) = 17.55 and P (12,I 197) = 11.04, respectively). In addition, the overall difference between the three was significant. All pairwlse contrasts on the time-of- day factor were also highly significant. The only significant interaction with the factor experiment was experiment-byquestion (F(4,1197) = 6.02, p < 0.0001). Furthr,r, the nine tests of quadratic trend in response times over days which were administered on the data collected for ,each question, for each time of day, all yielded slgdflcant results. The respective F ratios and p values are indicated in table I.

Several ape&c patterns in the data should be noted. Response times to the Tod quest&m are fastest within each part of day. The patterns associated with the Y and Tom questions, however, change with the time of day ln question. At noon they exhibit a reveml: in the beginn@ of the week responses to the first question are faster than msponser to the rtcond, but towards the end of the week the opposlte is

Page 5: Yesterday, today and tomorrow

413

Sun M 1 W Th F SOI DAYS OF THE WEEK

Fig. 3. Response times to Today, Yesterdav and Tomorrow questions (Evening).

the case. No such reversal is found in the other times of day, but the two show opposite patterns: in the morning responses to the Y question are faster than responses to the Tom question, whereas in the evening the opposite is the case.

Intro8pectioiu Several comments will be made on the introspective accounts given by the Ss.’

For all questions, and in all times of day, the frequency distribution of introspec- tions baaad on events in the day queried defined symmetric inverted U-functions. In contrast, the frequency distribution of introspections as!&iated with either the day preceding the one queried or the day following it were skewed. Introspections of the first type were dominant uftcr transitions from or to the weekend, whereas introspections of the second type were dominant before such transitions. The question-byday interaction in relative introspection frequency was highly signifi- cant @ < 0.005 or 0.001) for all times of day.

2 In order to avoid iengtby, cumbersome presentations, I avoid presenting the entire body of introspection distributions and the statistieal analyses carried on them. The distribution of htrospeetion for the Tod question is given in Appendix 1; the rest of the data may ba obtained upon request.

Page 6: Yesterday, today and tomorrow

474 B. Shanon/Yerter&y. to&y and tomowow

Errors In order to obtain 1260 correct responses, 1259 Ss had to be questioned. This

low rate and the lack of any dominant pattern in the error distribution excluded further analysis.

DiSCUSSiOtt

The discussion focuses on the three issues: replication, interpretation and extension. In terms of replication, that part of the present study which repeated the Koriat and Fischhoff study yielded a very good fit. Not only was the same inverted U-function found, but hideed the absolute latencies in the two studies are almost the same: in the Koriat and Fischhoff study response times ranged from 1.17 (Sunday) to 0.93 seconds (Friday) with a peak of 1.85 seconds (Wednesday); the respective values in the present study were 1.05 (Sunday), 1.26 (Satur- day) and 2.12 (Wednesday) seconds. The patterns of introspections in the two studies were also the same.

The various extensions of the present study, however, suggest that the interpretation given by Koriat and Fischhoff to their data is not valid. Koriat and Fischhoff claim that at the two ends of the week one determines what a day is on the basis of information having to do with the weekend. If this is the case then at the beginning of the week response times for the Y question should be shorter than those for the Tod question and both should be shcrter than the ones for the Tom question; the converse is expected at the end of the week: response times for the Tom question should be shorter than those for the Tod question and both should be shorter than the response times for the Y question. The data show a different pattern: in all parts of day response times to the Tod question were shorter than the ones associated with either the Y or the Tom questions. Further, in the noon data we note that at the beginning of the week subjects appeared to be faster in re- spondbg to the Y question than to the Tom one, whereas at the end of the week the opposite was the case. These results are puzzling because they indicate that the weekend does play a crucial role in the process of temporal orientation, yet not in the way suggested by Koriat and Fischlhoff.

In order to get out of this dilemma I propose an (admittedly, post- hoc) application of Collins and Loftus’ (1975) spreading activation model. The basic assumption is that in order to determine what day it

Page 7: Yesterday, today and tomorrow

B, ShanonlYastanW to&v and tomorrow 415

is, one consults any source of information which is available. One might refer to particular events on the day in question, to one’s knowledge about the previous or following day, to one’s memories and to one’s expectations - there is no guided search to each of these sources in particular, but rather a general activation of various sources of infor- mation. Such sources are more numerous at the beginning and end of the week than in the middle of the week. In fact, in the middle of the week the chances are that one is left only with information associated directly with the day in question” The variations in speed associated with al2 questions towards the weekend, then, reflect the high number of sources of information available.

Subjects’ introspections lend further support to the foregoing analy- sis. First, the distribution of introspections based on the day queried duplicates the inverted U-pattern found with response times. Second, in all parts of day, all responses to the Tod question were based in the middle of the week on associations with the day queried. In contrast, towards the weekends the percentage of such direct associations decreases and other types of associations appear. The changes, however, are not symmetric. Introspections associated with the day preceding the day queried are found after breaks in part of week, whereas ones associated with the day after it are found before such breaks.

So far, I have not referred to part of day variation. The introduction of this effort, however, was significant. As noted, in the morning responses to the Y question took less time than responses to the Tom question, whereas in the evening the opposite was the case. The follow- ing interpretation of this pattern is in line with the model sketched above. As the day progresses, information having to do with the previous day fades away and information having to do with the fo.llowing one becomes more prominent. Hence, in the morning it is more diffi- cult to respond to the Tom question than it is to the Y one, at noon the two are of comparable difficulty, and in the evening it is more difficult to respond to the Y question than it is to the Tom one. These results point out that the cognitive system is a dynamic one. Our data indicate that people do not determine temporal orientation by ‘flipping the pages of a mental calendar’ upon getting up in the morning. Rather, changes are continuously made all the time. Up to midday people still bear significant memories of the previous day, and it is already from midday on that the next day is taken into active consideration. In this perspective, it is interesting to note that responses to the Tod question

Page 8: Yesterday, today and tomorrow

476 8. Shivwa/Ymtedny, todsy d t-ow

were invariably the fastest. This suggests that people do ‘live in the present’. This significance of memories and expectations varies with the course of action people take, but the ‘now’ is always dominant.

Distribution of introspections for the Tod question in three times of day 4 $6).

Time wowti=G of&y bradon... h I Mon. lb. wed. mm. Fri. sat. Mom

Mm&g=> 20

ii

2 60 35 25 70 g 70 S 5; 61 28

Folbwiu~day 10 5 5 0 25 5 11

Noon ?kevhms&y 20 35 30 10 5 5 25 19 Dayquexiad 55 35 60 85 95 40 40 59

FoBouWdw 20 P8rtofweek 5 3: 0 5 0 20 10 0 0 35 3: 1:

Evenia8 Previous&y 20 50 45 20 5 15 50 30 Lkyquwiad 60 40 55 80 95 65 SO 63 F-day 20 10 0 0 0 20 0 7

References

Cdlins, A. W. and E. F. Loftus, 1975. A spteading activation theory of semantic process@. bychological Review 82.407428.

Rotiat, A. and B. Fischhoff. 1974. What day is today? An inquiry into the process of time orientation. Memory and Cognition 2.201-205.