some articles, and Erik Arthur for coding some moderators. The au- thors also thank ...... *Dinges, N. G., Tollefson, G., Parks, G. A., & Hollenbeck, A. R.. (1978).
Subject Embedding. Ungrammatical. Heavy NP Shift. The fact is irrelevant that John gave. Bill all his old books. Relative Shift. Plots have been hatched by many.
RICHARD A. BWCK. Montana State University, Bozeman, Montana and ... Bozeman, MT 59717 (e-mail: ..... termine whether the conditions shared a common effect size. ...... JSAS Catalog ofSelected Documents in Psychology, 10, 55. (Ms. No.
Concepts, Procedures, and Applications), and the 2001 spring issue of Music Per- ception (Vol. 18, No. 3). .... All of the herein presented measures typically follow some basic con- struction steps. First they transform ..... subjects had to indicate
However, Savitz (36) has postulated that, in general, the existence of a dose- ... (In German). In: Zabransky S, ed. SGA-Syndrom. ErnÃ¤hrung und Wachstum.
Judgments and Machine Learning to Match in Dynamic Environments. John P. Dickerson, Tuomas Sandholm. In AAAI, 2015. Presented by Zuxuan Wu ...
Mar 22, 2013 - mental representation described above (e.g., Shepard, 1958; Torg- erson .... signal-to-noise ratio of the estimates then depends on how closely.
Thomas M. Hess, Christina M. Leclerc, Elizabeth Swaim, and Sarah R. Weatherbee. North Carolina State University. It has been hypothesized that reductions in ...
A meta-analysis of prospective memory (PM) studies revealed that in laboratory settings .... issue relates to the predictions made by McDaniel and Einstein.
Scale and the Revised Wechsler Adult Intelligence Scale. (WAISâWAIS-R) method ..... November 2002. The resulting database includes 123 studies, contained in 104 articles of memory span and aging. Several studies contained more than one span task. T
memory span tasks, using graphical analysis on the original span metric. ..... for forward digit span, with R2 =.72; when we entered separate intercepts and ...
perceptual training holds great potential for remediation of a wide variety of performance deficits. These and ..... The module labeled STM represents short-term memory processes that serve ...... Journale Francais Ophthalmologie, 6, 473-479.
Mar 12, 2012 - biosynthesis pathway, nevertheless it has independent biological actions by .... melatonin receptor activation with signal transduction on tran-.
May 23, 2012 - Domenica Bueti, Centre Hospitalier ... tactile duration following exposure to pictures of threat, but not pictures of disgust ( ..... a âgoâ display which contained a central blue fixation dot (sub- ... ASSESSMENT OF EMOTIONS.
RESULTS: We found no significant difference between the annual odds of pertussis for the ... www.pediatrics.org/cgi/doi/10.1542/peds.2014-1729 ... grant funding, honoraria, and unrestricted educational funding from GlaxoSmithKline (GSK), Merck, Novar
M, Dabhadkar KC, Dahodwala N, De Leo D, Degenhardt L, Delossantos. A, Denenberg J, Des Jarlais ... De Motu Animalium. Gaesbeeck D, Boutesteyn C, de ...
Jan 17, 2015 - Brian K. Kennedy and Juniper K. Pennypacker. Over the last three ... increase in mouth ulcers at the high dose (20mg weekly). Importantly ...
These data show that if poultry processors harvest fil- lets earlier than usual (
Mar 29, 2006 - A.I. Mamedov is currently at the National Soil Erosion Research Lab., USDA-ARS- ..... mediate water contents (pF values of 3.3â2.8). Con-.
tion of dots, lines, or shapes attached to an unseen walking figure ... walking motion using only dots of light located at the joints ... trol the mental representation, connecting the dots with a stick ..... of human free-walking trajectories. In Pr
tion has declared it a global epidemic.1 Obesity in childhood is a cause of .... design and the possibility of this can be assessed by examining the methods of the ...
Predicting Human Similarity Judgments with Distributional Models: The Value of. Word Associations. Simon De Deyne1, Amy Perfors1 and Daniel J. Navarro2. 1 Computational ... formidable though they are, may not capture word representa- tions the way th
statistical theory and freely available software, that NLP can draw on. ... freely available R software, and it permits questions like the following to be asked: Can .... Pado et al. (2009) modified the system- level predictions approach to become â
Psychology and Aging 1998, \r>\. 13, No. 4, 584-596
Copyright 1998 by the Am
Human Aging and Duration Judgments: A Meta-Analytic Review Richard A. Block
Montana State University
Tel Aviv University
Peter A. Hancock Liberty Mutual Research Center
Differences in duration judgments made by younger and older adults were reviewed. Previous research is unclear about whether such differences exist and, if so, how they may be explained. The metaanalyses revealed substantial age-related differences. Older adults gave larger verbal estimates and made shorter productions of duration than did younger adults. There were no age-related differences in reproduction of duration or in psyche-physical slope relating judged and target duration. Older adults' duration estimates were more variable than were those of younger ones. Findings are discussed in terms of pacemaker rate and attentional resources. An explanation regarding divided attention between nontemporal and temporal information processing best explains the findings.
This implies that either many specific factors or a few general factors influence age-related differences. 2. Age-related influences on various cognitive variables are correlated. Furthermore, unique age-related influences are few and small. This implies that general factors must be postulated to account for the shared influences. 3. The speed of performance of simple cognitive tasks shares considerable age-related variance with many cognitive variables. This implies that factors related to simple processing speed or efficiency must be included in any explanation. For reasons we now explain, these generalizations suggest that older adults may differ from younger adults in timing and judging durations. Consider research and theories on this issue, which form the primary focus of our work.
Various cognitive processes change as a person ages. The slowing of information processing may be "the earliest, and most marked, symptom of cognitive aging'' (Wearden, Wearden, & Rabbitt, 1997, p. 962). Aging appears to be accompanied by a slowing of the component processes of many cognitive tasks, such as those that heavily involve working memory (Light, 1991; Salthouse, 1991), explicit memory retrieval (Craik & Jennings, 1992; Light, 1991), and divided attention (Salthouse, Regan, & Prill, 1984). There are conflicting theories on why age-related slowing may occur (Kausler, 1991). Slower processing speed and memory limitations may result from a reduction in general processing resources (Salthouse, 1994). Salthouse (1996) proposed some generalizations about cognitive aging, three of which are as follows: 1. Diverse cognitive variables reveal age-related differences.
Psychological Time and Aging Psychological time has been an important topic of research and theorizing for more than a century (Block, 1990; Michon & Jackson, 1985), perhaps because timing is essential to the optimal functioning of organisms. Psychological time involves processes by which an organism adapts to and represents the temporal properties of environmental events and relationships among them. Timing of durations ranging from seconds to minutes is essential for representing the immediate external environment. For example, while driving a vehicle or crossing a busy street, speed and time estimates provide vital information (Manser & Hancock, 1996). Many everyday behaviors involve short duration estimation, and it is important to understand the underlying processes and whether there are individual differences in them. Consequently, understanding how timing processes vary as a function of age is an important issue in its own right. Beginning with early essays on the psychology of time (Guyau, 1890/1988; Janet, 1877), several theorists have focused on the experience of duration. Early researchers have studied duration judgments for their intrinsic interest, such as in the context of psychophysical investigations (see Woodrow,
Richard A. Block, Department of Psychology, Montana State University; Dan Zakay, Department of Psychology, Tfel Aviv University, Tel Aviv, Israel; Peter A. Hancock, Liberty Mutual Research Center, Hopkinton, Massachusetts. Peter A. Hancock is now at Human Factors Research Laboratory, University of Minnesota. This study was supported in part by grants from the United StatesIsrael Binational Science Foundation and the MONTS (Montanans on a New Trac for Science) Program at Montana State University. Preliminary versions of this review were presented at the meetings of the Psychonomic Society, November 1994, and the International Society for the Study of Time, July 1995. We thank Robin Hall for helping to locate and retrieve articles, Erik Arthur for assisting with coding, Andrew Gilpin for providing the COEF\5\R program and discussing the coefficient of variation analysis, and Charles A. Pierce for discussing meta-analytic statistics. We also thank Fergus Craik, Alice Eagly, Anna Eisler, Hannes Eisler, Diane Halpern, Ralf Krampe, Charles A. Pierce, and Viviane Pouthas for helpful comments on a preliminary version of this article. Correspondence concerning this article should be addressed to Richard A. Block, Department of Psychology, Montana State University, Bozeman, Montana 59717-3440. Electronic mail may be sent to block® montana.edu.
AGING AND DURATION JUDGMENTS 1951). More recently, and arguably of more importance, researchers study duration judgment processes to clarify more general physiological, memorial, and attentional processes. Here we also focus on duration judgments. Early theorists speculated about possible differences in duration judgments between younger and older adults. Janet (1877) proposed that the ratio between a duration and a person's age influences the person's duration judgments. James (1890) claimed that' 'the same space of time seems shorter as we grow older," but he Qualified this by saying "that is, the days, the months, and the years do so; whether the hours do so is doubtful, and the minutes and seconds to all appearances remain the same" (p. 284). Mach (1900) also thought that the physiological unit of time lengthened with age. A controversy ensued after the publication of an article by Nitardy (1943) that essentially echoed these previous comments. Authors of nine additional short articles (e.g., Carlson, 1943) replied to Nitardy, but they provided no empirical data. Older adults do indeed report that time passes quickly (Baum, Boxley, & Sokolowski, 1984) and at a faster rate than when they were younger (Schroots & Birren, 1990). Research using questionnaires tends to support this view (Joubert, 1983; Lemlich, 1975; Walker, 1977). Cohen (1967) speculated about why the years seem to pass more quickly as a person ages. He attributed this, in part, to a slower metabolic rate. He thought this slowing would have an effect on duration judgments similar to the hypothetical effect of reduced body temperature, that is, subjective time would decrease and the calendar year would seem shorter. Fraisse (1957/1963) suggested that "decreased biological activity may make us register fewer changes" (p. 248), thereby leading to an apparent speeding in the rate of physical time. However, he cautioned that older adults may compensate for any such influences, claiming (on the basis of little evidence) that there is "hardly any change in their objective timejudgments" (Fraisse, 1957/1963, p. 248). Whitrow (1972) also proposed that "our sense of temporal duration . . . depends on our age, for our organic processes tend to slow down as we grow older, so that, compared with them, physical time appears to go faster" (p. 43). More recently, Craik and Hay (in press) also suggested that the reason why older adults perceive time as passing at a relatively fast rate is that a biological clock slows progressively as a person ages. The slowing of a biological clock would presumably give rise to the feeling that external time passes faster. Experiments comparing the magnitude of duration judgments made by older and younger adults date from the late 1950s (Feifel, 1957; Goldstone, Boardman, & Lhamon, 1958). Nevertheless, we know of no systematic description or understanding of age-related differences in duration judgments that has emerged, and we know of no comprehensive and recent review of experimental findings that has been published. Models of Duration Judgment Processes Theorists have proposed several kinds of models to explain duration judgments (Block, 1990). One kind emphasizes physiological mechanisms. In the most common variant, an internal clock, consisting of a pacemaker, an accumulator, a reference memory, and other components, subserves time-related behavior (Wearden & Penton-Voak, 1995). The pacemaker rate may be
influenced by biological variables such as metabolism and brain temperature, neurotransmitter receptor sites, psychoactive drugs, and arousal level. Because basal metabolism and brain temperature decrease as a person ages (Altman & Dittmer, 1968; Kadlub. 1996), the pacemaker rate may also decrease. A person may compensate continuously (recalibrate) for this hypothetical slowing, however, by changing the contents of the reference memory store, so that fewer pacemaker-produced pulses are needed to correspond to a learned duration unit (such as 1 min). Another kind of model proposes that duration is a cognitive construction that is influenced by attention and memory processes (Block, 1990). Several kinds of methodological and individual factors may influence duration judgments because of their effects on attention and memory (Block, 1985, 1989, 1990). In light of these possibilities, the choice of a method to investigate age-related differences in duration judgments may be critical. Whether duration judgments show any age-related differences may depend on the specific methodology used to obtain them.
Methodological Factors Duration Judgment Paradigm, One important variable influencing duration judgments is whether a person knows in advance that a duration estimate will be required. In the prospective paradigm, a person has this knowledge; in the retrospective paradigm, a person does not. The magnitude and Intel-individual variability of duration judgments depend critically on this difference in duration judgment paradigm (Block & Zakay, 1997). However, we were not able to assess age-related differences in retrospective duration judgments or investigate duration judgment paradigm as a moderator variable. The only articles we know of using the retrospective paradigm revealed methodological problems (Kelley, 1980) or contained insufficient statistics (Vanneste & Pouthas, 1995) and yielded no data relevant to our meta-analyses.
Duration Judgment Method Another important variable is the method used to obtain duration judgments. Researchers typically use one of three main methods: verbal estimation, production, and reproduction. These methods entail somewhat different cognitive processes, and the resulting duration judgments reveal characteristics of those processes (Zakay, 1990, 1993; Zakay & Block, 1997). The methods of verbal estimation and production involve comparing duration experience with internal (reference memory) information concerning conventional duration units such as seconds and minutes. In the method of verbal estimation, a person experiences a target duration and then must translate his or her experience of it into an estimate in clock units, usually seconds. This translation introduces variability into the judgments because different people have learned to accomplish this with different proficiency. In the method of production, a person attempts to delimit an objectively measured duration corresponding to a verbally stated time period (e.g., "Say stop after what seems like 60 s to you"). As with verbal estimates, this kind of judgment requires a translation, but in this case it is in
BLOCK, ZAKAY, AND HANCOCK
the reverse direction (i.e., from an objectively labeled duration to a subjectively experienced duration). In a variant on this method, the method of repeated production, a person delimits several objectively labeled durations of a specified length, usually 1 s (e.g., "Press this button once every second until I tell you to stop''). Verbal estimation and production methods have drawbacks in some experimental contexts because of the need to assume that the translation between conventional units and experienced duration is reliable. These methods, however, may be suitable to investigate individual differences (or effects of variables that may influence the rate of internal processes). Researchers have successfully used them in studies of manipulations thought to influence the rate of internal time-keeping processes, such as drugs (e.g., Frankenhaeuser, 1959; R. E. Hicks, 1992). However, if internal clock or time-keeping processes change gradually with age, a person may use feedback to recalibrate internal time keeping against physical units of duration (Surwillo, 1968). The method of reproduction does not involve verbally stated units. A person experiences a target duration and then attempts to delimit another time period that is the same length. This method, which relies on a comparison of experiences instead of on conventional duration units, has the potential drawback that reproductions may be "an index only of the consistency of the subjective time base; [they provide] no information as to the rate itself" (Cahoon, 1969, p. 261). Even if the rate of physiological and cognitive processes varies with age, the same rate will subserve a person's experiencing the target duration and reproducing it. Thus, the reproduction method may detect individual differences only if it is used in the framework of psychophysical studies, in which duration is varied. In addition, judgments obtained by using the reproduction method (as well as the production method) may be confounded by extraneous variables such as the desire to terminate the experiment sooner, impatience, or the inability to delay a response.
Age-Related Differences Individual differences in duration judgments may originate in various physiological parameters and cognitive processes, including (a) the rate of biological processes, such as those thought to underlie an internal pacemaker; (b) the basal metabolic rate or brain temperature, perhaps by influencing the pacemaker rate; (c) the strength, clarity, duration, or variability of memory processes, such as those that mediate the encoding of information during a duration; and (d) attentional resources or resource allocation, especially involving attention to time. In our review, we focus mainly on the two hypotheses that relate most directly to the available evidence: the pacemaker-rate hypothesis and the attention hypothesis. (Later, we also discuss the metabolism or temperature hypothesis and the memory hypothesis.) An attentional-gate model (Zakay & Block, 1997), which combines pacemaker and attentional components in an integrated model, emphasizes that duration timing may require a person to divide attention between nontemporal (stimulus) information processing and temporal information processing. The magnitude of prospective duration judgments is mainly influenced by the amount of attention a person allocates to pro-
cessing temporal information. The attentional-gate model can account for much of the evidence (Zakay & Block, 1997). Assuming that there is an age-related slowing of a pacemaker rate, a decrease in attentional resources, or both, there are two possible scenarios, neither of which is currently supported by evidence. First, older adults may learn to compensate for the slowed pacemaker or decreased resources. In this scenario, one expects no age-related difference in duration judgments. Alternatively, older adults may not sufficiently recalibrate or compensate for the slowed pacemaker or decreased resources. In this scenario, older adults will give smaller verbal estimates and make longer productions of durations than will younger adults. However, an age-related difference in pacemaker rate or attentional resources should not influence reproductions because the method of reproduction involves a comparison of durations (the standard, or target, duration and the comparison, or reproduced, duration). Thus, both pacemaker and attentional models (e.g., Zakay & Block, 1997) predict that duration judgment method will be an important moderator of age-related differences in duration judgments, if any difference is found. One crucial distinction between pacemaker and attentional hypotheses is that pacemakers are assumed to operate fairly autonomously, perhaps being influenced by biological variables but not by cognitive factors. However, attentional models make specific predictions concerning cognitive factors such as information-processing difficulty. Prospective duration timing requires a person to divide attention between nontemporal (stimulus) information processing and temporal information processing (Brown, 1997; Zakay & Block, 1997). Researchers have not yet found the exact nature of the cognitive processes that occur when a person is said to attend to time. Indeed, such processes may vary depending on the person and the situation (e.g., duration judgment method). Attending to time may involve judging the recency of the signal that indicated the start of a time period, engaging in chronometric counting, imagining events signaling the end of a time period, or a combination of these and other such processes. Although research has indicated that attending to temporal information and attending to nontemporal information compete for general attentional resources, the extent to which this implies a working-memory limitation, or a limited capacity within some functional subsystem of working memory, remains unclear. The attentional-gate model of prospective duration judgment somewhat clarifies the notion of attending to time (Zakay & Block, 1997). In this model, attending to time means focusing on the stream of signals produced by a pacemaker and transferring a representation of the total number of signals to a working-memory store. Thus, if a person allocates relatively more attention to time, more signals are included in the working-memory representation, and the expected result is that a person will give relatively larger verbal estimates and make relatively shorter productions of duration. Duration judgment evidence may therefore test the notion that divided-attention tasks affect older adults in a relatively severe way, whereas tasks not requiring divided attention may not affect older adults any more than they do younger adults. If this were the case, attentional models predict that the difficulty of the nontemporal information-processing task will moderate any age-related effect on prospective duration judgments (cf. Craik & Hay, in press). There may be little or no age-related
AGING AND DURATION JUDGMENTS
difference if the primary (nontemporal) task is easy. However, as nontemporal task difficulty increases, older adults may accumulate relatively less temporal information than would younger adults. The consequences of this on duration judgment magnitude will depend on the specific method used. If the primary (nontemporal) task is more difficult, younger adults may still have enough attentional resources to allocate to the secondary (temporal) task, but older adults may have relatively few residual resources. As a result, older adults would give relatively smaller verbal estimates and make relatively longer productions. The effect on reproductions would depend on whether the target duration and the reproduction duration are filled with equally difficult tasks. If they are, no age-related difference is expected; if, instead, the target duration is filled with a difficult task and the reproduction duration is empty (as is usually the case), older adults' reproductions should be relatively short (just as their verbal estimates are smaller). In summary, studying agerelated differences in duration timing may provide crucial converging evidence on the impact of aging on cognitive processes, such as those involving attention and working memory.
an internal pacemaker because if the pacemaker rate decreases (and there is no compensation for this change), the psychophysical slope may also decrease. In other words, age and duration length may interact to influence duration judgments. In summary, in our meta-analytic review we examined evidence for age-related changes in the magnitude and variability of duration judgments. Second, we clarified theories on changes observed in cognitive aging. Finally, we addressed the general issue of individual differences in temporal experience and judgment, which remains one of the most intriguing and poorly explained issues in the psychology of time.
Method Sample of Studies We searched a database containing more than 9,000 references on the psychology of time (Block & Eisler, 1998). It includes articles from the following
Abstracts (1923-1966) and
PsyclNFO (1967-1997), using the keywords time perception and time estimation; vvn&Medline (1966-1997), using the keyword time perception} It also includes references from published bibliographies on time research: references from articles, books, and book chapters; and refer-
Specific Goals of the Present Meta-Analyses The goal of our meta-analytic review was to evaluate evidence on age-related differences in human duration judgments while also discovering what variables (e.g., duration judgment method) moderate any obtained differences. The outcome may clarify or suggest theories concerning processes that produce age-related differences. The finding of moderator variables is vital information that clarifies proposed models. (In a metaanalysis, the finding of a moderator variable is analogous to the finding of an interaction effect in primary research.) Studies investigating duration judgments almost always report the duration length used, so we also accumulated primary-level statistics based on a common scale of measurement, the ratio of subjective to objective duration. In this regard, the present review differs from most meta-analytic reviews: The primary-level statistics clarify the meta-analytic statistics. Several articles on age-related differences contain data on the interindividual variability of duration judgments, such as standard deviations. (None reported any data on intraindividual variability, another potentially interesting measure.) In spite of this, we know of no researchers who reported inferential statistics on variability data, and few commented on differences in variability. Because interindividual variability is theoretically relevant, we reviewed these data. Thus, the present meta-analysis reveals new information about interindividual variability in duration judgments, thereby suggesting whether different participants use similar or different kinds of processes (Wearden et al., 1997). Greater interindividual variability is expected if different participants use different processes and these processes affect duration judgments differently. It may also be expected if different participants use the same (or similar) processes, but the intraindividual variability of those processes differs (especially if only a few duration judgments are obtained). Finally, we evaluated several studies in which the slope of the psychophysical or regression function relating subjective and objective duration was reported. These data may provide a relatively direct measure of hypothetical changes in the rate of
ences from our files (see Block & Zakay, 1997). We also searched Social Sciences Citation Index (Social SciSearch, 1977-1997) for articles that cited major articles, such as those by LeBlanc (1966, 1969). Finally, we checked the reference lists of all relevant studies. To be included in our meta-analyses, an experimenter must have studied normal human participants of diverse ages, and a published article must have contained quantitative data on duration judgment magnitude in more than one age category (young adult, middle-age adult, and old adult). We compared younger adults and older adults, for whom younger and older participants' ages were coded variables. This process necessarily excluded any study in which age varied only within the narrow range of a single category (e.g., Fitzpatrick & Donovan, 1978; Grossman & Hallenbeck, 1965: Mentzer & Schorr, 1986; Newman & Gaudiano, 1984). It also excluded any study in which researchers did not obtain and report duration judgment magnitude, specifically (a) studies in which no experimental duration was presented or estimated, including those obtaining only questionnaire data (e.g., Baum et al., 1984; Gallant, Fidler, & Dawson, 1991; Joubert, 1983; Kuhlen & Monge, 1968; Lemlich, 1975; Walker, 1977); (b) experiments in which participants judged other temporal dimensions, such as recency, or lag (e.g., McCormack, 1982); (c) experiments in which participants received feedback on their judgments, thereby learning to make them more accurate (e.g., Arenberg, 1968; Lejeune & Pouthas, 1991; Rammsayer, Lima, & Vogel, 1993); (d) experiments in which participants made qualitative (e.g., same or different, or shorter or longer) judgments, such as duration-discrimination judgments, with only discrimination threshold data reported2 (e.g., Arenberg, 1968; Goldstone, Lhamon, & Boardman, 1957; Rammsayer et al., 1993; Smythe & Goldstone, 1957); (e) experiments in which researchers reported only data on duration judgment error, percentage of correct judgments, or percentage underestimation or overestimation (e.g., Arenberg, 1968; Bell, 1972; Lejeune & Pouthas, 1991; PumpianMindlin, 1935); and ( f ) experiments that might not have been excluded except that researchers did not report sufficient statistics (e.g., mean
We included any relevant study listed in PsyclNFO or Medline as of December 1997. 2
Without assuming a mathematical model, this kind of judgment can-
not be compared with quantitative duration judgments of the kind metaanalyzed here. Including them would subject the meta-analysis to an "apples and oranges" criticism (Sharpe, 1997).
BLOCK, ZAKAY, AND HANCOCK
judgment, numbers of participants, or inferential statistics) to estimate
The homogeneity of each set of ds was tested with the Q statistic to
an effect size or the duration judgment ratio (e.g., Dmitriyev, 1980; R.
determine whether the conditions shared a common effect size. If there
A. Hicks, Bramble, &Ulseth, 1967; Kline &Burdick, 1980; Vanneste&
was significant heterogeneity of effect sizes, we attempted to account
Pouthas, 1995). We also excluded unpublished dissertations (Bull, 1973;
for it with coded or manipulated study attributes. Two coded variables,
Kelley, 1980) and conference presentations,3
publication year and the total number of duration judgments, were continuous. Those variables were fit with a weighted least-squares regres-
sion model (Hedges, 1982b; Hedges & Olkin, 1985) using SPSS and
To identify theoretically important moderator variables, we coded the
DSTAT (Johnson, 1989,1993). All other coded attributes were categorical. For these, we used categorical models (Hedges, 1982a; Hedges &
following variables from each experiment and each within-experiment
Olkin, 1985), as implemented by DSTAT. (We combined two similar
condition: (a) publication year; (b) participants' sex (female, male,
classes of a variable if there were fewer than three effect size estimates
both male and female, or unknown); (c) participants' age (young adult, such as most samples involving college students [18.0-29.9 years of
in a given class.) These techniques yield a between-classes effect, which
age], middle-age adult [30.0-59.9 years of age], or old adult [at least
also calculated correlations among coded variables to determine the extent to which they were independent.
60 years of age]); (d) predominant duration length (very short [4.9 s
reveals whether that variable significantly moderates the age effect. We
or shorter], short [5.0-14.9 s], moderate [15.0-59.9 s], or long [60.0 s or longer]); (e) duration judgment method (verbal estimation, produc-
tion, repeated production, or reproduction); and ( f ) total number of duration judgments made by each participant during the experiment.
For each condition analyzed, we calculated the ratio of subjective to
Two of us coded study attributes independently, resolving any disagree-
objective duration—hereinafter called the duration judgment ratio—
ments by discussion.
separately for each age class. This is a standard measure calculated and
We also coded other theoretically relevant variables, but, for these,
reported in many duration judgment studies (Hornstein & Rotter, 1969).
only one class of the variable was adequately represented; the others
It enables a comparison of duration judgments across conditions and
contained fewer than three effect size estimates. No conclusions may be drawn for such variables. Future research is needed to reveal whether
experiments that used different durations. It reverses the commonly
any of these variables moderate age effects on duration judgments. All
methods (primarily verbal estimation) because the participant's produc-
(or nearly all) experiments (a) used a prospective duration judgment paradigm; (b) did not state whether participants' watches were removed;
tion is the actual (objective) duration corresponding to the verbally
(c) did not manipulate participants' body temperature; (d) did not use
influence of duration judgment method without an artifactual influence
any arousal manipulation; (e) presented either no stimuli or one simple,
of the typical negative correlation. We also calculated the mean ratio of
continuous stimulus; ( f ) did not segment the duration with salient mark-
older-to-younger mean duration judgment ratios, hereinafter called the
ers or high-priority events; (g) had few environmental or background
age ratio. Accumulating these primary-level statistics across conditions
changes; (h) required only passive or covert and easy or shallow pro-
and experiments clarifies die meta-analytic statistics (Block & Pierce,
cessing of any presented stimuli; (i) did not involve chronometric count-
1998).5 Two-tailed / tests were performed on these unweighted primary-
ing; (j) used no delay, except possibly for brief instructions, preceding
level statistics. Our descriptions of results take into account whether a
the duration judgment; (k) did not require memory for presented infor-
particular comparison was significant (p < .05).
mation; (1) had no changes in type or level of processing; and (m) had no concurrent task.
found negative correlation between the production method and the other
requested (subjective) duration. Thus, this ratio assesses the moderating
We also analyzed separately the experiments that provided sufficient information, such as standard deviations, to calculate the relative interindividual variability of duration judgments made by participants of differ-
ent ages. One cannot simply compare standard deviations because they typically increase with increasing mean judgment when a ratio scale of
Two of us independently estimated effect sizes, resolving any dis-
measurement is involved. We instead used the common psychometric
agreements by discussion. Each effect size was calculated as g, the
measure, the coefficient of variation, which is the standard deviation
difference between the mean duration judgment given by participants of
divided by the mean judgment (see, e.g., Morse, 1993; Wearden et
two age classes divided by the pooled standard deviation (Hedges &
ah, 1997). This makes the reasonable assumption that duration timing
Olkin, 1985), using the computer software DSTAT (Johnson, 1989,
involves a Poisson process. The program COEFVAR (Gilpin, 1993) was
1993). Each g was then converted to d by correcting it for bias, weighting
used to calculate a chi-square value for the difference between coeffi-
by the reciprocal of its variance, TW (Hedges, 1981; Hedges & Olkin,
cients of variation with the Bennett-Shafer-Sullivan likelihood ratio
1985). The ds were combined by separately calculating unweighted and
test (Shafer & Sullivan, 1986). DSTAT was then used to convert each
weighted means.4 If a researcher manipulated a potential moderator variable (e.g., used different duration judgment methods) and provided adequate informa-
Meta-analysts do not agree on whether one should include unpub-
tion to calculate separate effect size estimates for each level of the variable, we did so for that moderator analysis. (Tb provide a single
lished data (Sharpe, 1997). When we included Bull's (1973) data, our
estimate of g for each experiment, we averaged all such separately calculated effect sizes.) Thus, each moderator analysis contained mainly
Tables 1 and 3). Kelley's (1980) data, however, were excluded for
findings and conclusions were not altered in any substantial way (see
study effect sizes as well as a few within-study effect sizes. Using more
methodological reasons. 4 See Block and Zakay (1997) for a more complete description of
than one effect size estimate from the same experiment violates the
the procedures used.
assumption that the effect sizes are independent. However, this violation
We also calculated weighted mean ratios, weighting each duration
apparently does not substantially affect statistical precision (Tracz, 1985;
judgment ratio (and the age ratio) by the sample size involved. This is
Tracz, Elmore, & Pohlmann, 1992). If we had not calculated separate effect size estimates for each level of a manipulated variable, we could
similar to die weighting (by TW) that is involved in using Js as effect size estimates in the meta-analyses. Although doing so often increased
not have properly conducted the moderator analyses because we would have excluded (or coded into a mixed category) some of the most
the magnitude of the age-related differences reported here, it did not alter any conclusions. We report here the unweighted mean ratios and
results of t tests based on them.
AGING AND DURATION JUDGMENTS chi-square to an effect size. We also accumulated and tested primarylevel statistics on coefficients of variation. As noted earlier, (see p. 587), we also evaluated studies that reported data concerning the slope, or exponent (/9), of the psychophysical or regression function relating subjective and objective duration across different duration lengths.
if the mean duration judgment ratio was greater for older participants and as negative if it was greater for younger participants. The resulting weighted mean effect size, d+ = 0.40, 95% confidence interval (CI) = 0.28 to 0.52, indicated that the duration judgment ratio was greater for older than younger participants (p < .0001). Effect sires were not homogenous, Q(13) = 31.5, p = .003, so we used coded study attributes to account for
Results and Discussion
variability in them.
A total of 16 experiments, published in 16 separate journal articles, met all criteria for inclusion in our review. A total of 15 articles were written in English and 1 in Spanish. The median publication year was 1977.5. The mean age of participants was
The d+ of 0.40 is considered small to moderate in magnitude. The mean duration judgment ratio was greater for older than for younger participants, r(13) = 4.25, p < .001, and the mean older-to-younger (age) ratio of 1.21 was greater than 1.00, t( 13)
about 25.1 for young adults, 43.2 for middle-age adults, and
= 24.5, p < .001. Even though the effect size was only small
70.8 for old adults.
to moderate, the age ratio revealed a relatively large difference: The mean duration judgment ratio was 21% greater for older than for younger participants, which is in the opposite direction
Duration Judgment Magnitude
from what is usually predicted.
Size and Primary Statistics
Of the 16 experiments, 14 contributed an effect size for mag-
nitude of duration judgments made by younger adults compared with older adults (see Table 1). All used a between-subjects
Table 2 shows the results of model testing involving the only
design; none was a longitudinal study. A total of 12 effect sizes
significant categorical moderator variable: duration judgment
were calculated from means and standard deviations, standard
method. Duration judgment method. Because only two experiments used the method of repeated production, we combined them
errors, or similar measures of variability; 1 from means and a pooled standard deviation estimated from a related F value; and 1 from a reported F value. We denned an effect as being positive
with the seven that used the method of production. Duration
Table 1 Duration Judgment Ratios and Effect
Feifel (1957)* Goldstone et al. (1958rb Surwillo (1964)=-'' McGrath & O'Hanlon (1968)'-' McNanamy (1968)'* LeBlanc (1969)'-" Neuringer & Levenson (1972)'* Bull (1973)""-1 Newman (1976; see 1982)' Kline et al. (1980)' Landaeta et al. (1981)' Licht et al. (1985)' Beck (1988)' Polyukhov (1989)"'" Eisler & Eisler (1994)" Overall mean (unweighted^ Overall mean (weighted)*
Note. Position effect size (d) indicates that the mean duration judgment ratio (subjective-to-objective duration) was greater for older adults; negative effect size (d) indicates that it was greater for younger adults. Older = middle-age or old adults, hunger = young or middle-age adults. * An approximate datum (e.g-, one estimated from a figure). a Study compared old adults and young adults. b We excluded Smythe and Goldstone's (1957) "passive estimation" data. c Study compared old adults and middle-age adults. d We pooled data from two subgroups. e Study compared middle-age adults and young adults. f We used a median split for age (20-30 and 30-50 years old, respectively). g We combined pretest and trials. * We included only geriatric and normal groups. ' We show these unpublished data for the sake of completeness; we did not include them in any of the analyses. ' Each mean weights each experiment equally. k Each mean weights by n contributing to each datum (and by TW for effect size).
BLOCK, ZAKAY, AND HANCOCK
Test of Categorical Model for Duration Judgment Comparisons 95% CI for d,+
Variable and class
Duration judgment method Verbal estimation Production or repeated production Reproduction
Homogeneity within class
0.27 0.60 -0.02
0.05 0.44 -0.31
0.48 0.76 0.27
3.58 13.56 6.20
No, Note. Positive mean effect size (rfH-) indicates that the mean duration judgment ratio (subjective-to-objective duration), was dun ... =greater for older adults; negative mean effect size (Jj_) indicates that it was greater for younger adults. a None of the values was sufficiently large to reject the hypothesis of homogeneity of effect sizes »*;> < .001.
judgment method moderated effect sizes. The age effect was greater for the method of production or repeated production than for the methods of verbal estimation (p < .05) and reproduction (p < .001). Only effect sizes for production and verbal estimation were significantly greater than zero. No method category showed significant heterogeneity of variance, so duration judgment method adequately accounted for the age effect. Older participants gave a greater duration judgment ratio in conditions using the methods of verbal estimation (1.42) and production or repeated production (1.36) than in conditions using the method of reproduction (0.95). However, younger participants gave a greater duration judgment ratio using the method of verbal estimation (1.19) than the method of production or repeated production (1.05) and the method of reproduction (0.91). The age ratio was greater for production or repeated production (1.28) and verbal estimation (1.19) than for reproduction (0.99); the former two did not differ. Figure 1 clarifies the significant moderating effect: Compared with younger adults, older adults gave larger verbal estimates and made shorter productions (both of which entail a greater duration judgment ratio), but they made comparable reproductions. The findings concerning verbal estimates and productions are the opposite of what some theories predict, such as the notion that a pacemaker rate slows with aging. The conclusion that reproductions do not show age effects should be regarded skeptically because, although Eisler and Eisler (1994) reported small effects, they were significant. Publication year. The moderating influence of publication year was significant (Q R = 5.60, p = .01); however, the linear regression model did not provide a good fit (GE = 25.47, p .01). The correlation between publication year and effect size was marginally significant, r(12) = —.50, p = .07, and the slope of the regression function was -.44. Thus, earlier studies showed a greater effect size. This may be explained by the fact that experiments using the method of production, which yielded the largest age effects, tended to be published earlier (median date = 1969) than those using the methods of verbal estimation and reproduction (median dates = 1984.5 and 1981, respectively). Other variables. The median number of duration judgments that each participant made in an experiment was four. The linear
moderating influence of number of duration judgments was not significant (Qs = 1.62, p = .20), and the linear regression model did not provide a good fit (QE = 29.84, p < .003). Several other variables were sufficiently represented across studies or frequently manipulated in experiments. However, there was no significant moderating influence of older participants' age, younger participants' age, or participants' sex. (See Tables 1,3, and 4 for information about the age classes of participants in each experiment.) We describe the relevant findings concerning duration length here because of their theoretical importance (also see the follow-
• Verbal Estimation
co DC c
E en •a
1.25 c o
Mean Age Figure 1.
Mean duration judgment ratio as a function of age class
shown separately for each duration judgment method.
AGING AND DURATION JUDGMENTS
Table 3 Coefficients
of Variation (CV) and Effect
Older Study aJ>
Goldstone et al. (1958) Surwttlo (1964)"1 LeBlanc (1969)"" Neuringer & Levenson (1972)Bj Bull (1973)"-"-8 Licht et al. (1985)c Beck (1988)' Polyukhov (1989)1-" Overall mean (unweighted)11 Overall mean (weighted)'
Note. Positive effect size (d) indicates that the mean coefficient of variation was larger for older adults. Older = middle-age or old adults. 'Vfounger = young or middle-age adults. * An approximate datum (e.g., one estimated from a figure). * Study compared old adults and young adults. b We excluded Smythe and Goldstone's (1957) "passive estimation" data. c Study compared old adults and middle-age adults. d We pooled data from two subgroups. ' Study compared middle-age adults and young adults. 'We included only geriatric and normal groups. B We show these unpublished data for the sake of completeness; we did not include them in any of the analyses. h Each mean weights each experiment equally. ' Each mean weights by n contributing to each datum (or by TW for effect size).
ing section on psychophysical slope). Across all experiments, the target duration ranged from 1.3 s to 480.0 s.6 Duration length was not a significant categorical moderator variable, &(2) = 2.48, p = .29, and all weighted mean effect sizes were significantly greater than zero (d+s = 0.34, 0.34, and 0.53 for very short and short durations, moderate durations, and long durations, respectively).1
Seven published reports contained enough information to calculate coefficients of variation for older and younger adults (see Table 3). The sign of each effect size was positive if the mean coefficient of variation was greater for older participants and was negative if it was greater for younger participants. The resulting weighted mean effect size, rf+ = 0.36, 95% CI = 0.20 to 0.52, indicated a greater coefficient of variation for older than for younger participants (p < .0001). The homogeneity of effect sizes was indicated, Q(6) = 1.13, p = .98, and no variable significantly moderated them. The weighted mean effect size, rf+ = 0.36, although significant, is considered small to moderate in magnitude. The mean coefficient of variation was greater for older than younger participants, r(6) = 6.08, p < .001, and the mean age ratio of 1.39 was greater than 1.00, t(6) = 20.7, p < .001. Even though the effect size was small to moderate, the age ratio revealed a relatively large difference: The mean coefficient of variation was 39% greater for older than younger participants, indicating that older adults showed substantially more interindividual variation than did younger adults (see Figure 2).
duration and target duration for older and younger adults (see Table 4). In two experiments, Salthouse, Wight, and Ellis (1979) obtained analog magnitude judgments of very short (401,000 ms) durations of light or dark. Nichelli, Venneri, Molinari, Tavani, and Grafman (1993) obtained verbal estimates of a 5to 40-s task (exposing digits at a 1-s rate). Eisler and Eisler (1994), who performed the most extensive study, had participants reproduce 1.3- to 20.0-s tones. The sign of each effect size was positive if the slope was greater for older participants and was negative if it was greater for younger participants. The resulting weighted mean effect size, d+ = 0.04, 95% CI = —0.29 to 0.37, indicated no significant difference in slopes for older and younger participants (p = .82). The homogeneity of effect sizes was indicated, Q(3) = 0.37, p = .95. Thus, there was no significant age-related difference in the slope of the function relating duration judgments to actual duration. These data are inconsistent with the notion that there is an age-related difference in pacemaker rate that is not recalibrated (compensated) by a gradual learning of correspondences between pacemaker pulses and numerical labels for durations. In other words, if there are age-related differences in pacemaker rate, these differences are also at least adequately compensated for by changes in reference memory information.
Summary and Theoretical Discussion The duration judgment meta-analyses revealed several important findings regarding the influence of age on mean duration
'Additionally, Salthouse, Wright, and Ellis (1979) used very short
Psychophysical Slope Four published experiments provided data on the slope of the psychophysical or linear regression function relating judged
(40-1,000 ms) durations; see the section on psychophysical slope. 7
Duration length was also tested as a continuous moderator variable
using the mean target duration when necessary. The linear moderating influence of duration length was not significant (Q, = 0.12, p = .73).
BLOCK, ZAKAY, AND HANCOCK
the only important moderator variable. Older adults differed from younger adults in that they gave larger verbal estimates and made shorter productions of duration, but they made fairly comparable reproductions. This was the case regardless of the duration length used. Considerable evidence reveals that judgments of durations less than several seconds entail different processes than judgments of longer durations (for reviews, see Block, 1979, 1990; Fraisse, 1984; Poppel, 1985/1988). However, our age-related differences in duration judgments did not depend on duration length. The findings concerning verbal estimates and productions were the opposite of what some theories predict, such as the notion that a pacemaker rate slows with aging. For these methods, older participants' judgment ratios (relating subjective duration to target duration) were greater, not less, than those of younger participants. As McGrath and O'Hanlon (1968) aptly noted, "a given interval of time seems longer as [a person! grows older. This is in direct opposition to the traditional view" (p. 1087). Licht, Morganti, Nehrke, and Heitnan (1985) commented that
Verbal Estimation Production
O C OS ID
5 0.20 -
conflicting interpretations questioning whether brief intervals of real time are perceived as passing more or less rapidly with increasing age have been reported. Reasons for these inconsistent results lie in semantic confusions, in variations in methodology, and in the effects of other mediating [moderating] variables, (p. 211)
Mean Age Figure 2.
Mean coefficient of variation as a function of age class
shown separately for each duration judgment method. No experimental report contained enough data to calculate coefficients of variation for the reproduction method.
judgment. The overall weighted mean effect size suggested that older participants' duration judgments differed significantly from those of younger participants: The ratio of subjective to objective duration was relatively larger for older adults. When we tested categorical models investigating potential moderator variables and aggregated primary-level statistics (duration judgment ratios), the duration judgment method was found to be
Table 4 Psychophysical or Regression Slopes and Effect Older
Our analysis suggests one important moderating variable: Agerelated effects may be found using only absolute methods, verbal estimation or production, in which a person must translate from subjective experience to a conventional verbal label (i.e., numerical quantity), or vice versa. There is not yet enough evidence that older adults differ from younger adults on the most commonly used duration judgment method that requires a comparison of durations: reproduction. The mean coefficient of variation was greater for older than for younger adults. As we mentioned earlier, there are several possible explanations for this finding, such as greater interindividual variation in the kinds of processes or greater intraindivid-
Salthouse et al. (1979, Experiment 1)" Salthouse et al. (1979, Experiment 2)' Nichelli et al. (1993)" Eisler & Eisler (1994)E Overall mean (unweighted)11 Overall mean (weighted)'
1.31 1.23 0.90* 0.87 1.08* 1.08*
18 1.33 22 1.21 15 0.83*
0.87 1.06* 1.09*
0.98 0.98 1.08* 1.00 1.02* 0.99*
33 45 20 48 4 146
-0.06 0.06 0.29* 0.01 0.08* 0.04*
Note. Positive effect size (d) indicates that the psychophysical or regression slope was larger for older adults; negative effect size (d) indicates that it was larger for younger adults. Older = middle-age or old adults, hunger = young or middle-age adults. * An approximate datum (one estimated from a figure). * Study compared old adults and young adults. b Study compared old adults and middle-age adults. c Study compared middle-age adults and young adults. d Each mean weights each experiment equally. c Each mean weights by n contributing to each datum (or by TW for effect size).
AGING AND DURATION JUDGMENTS
ual variation in common component processes (see also Morse, 1993). Only a few experimental reports contained data on the slope of the psychophysical or regression function relating judged duration to target duration. There were no significant age-related differences in the slope, a finding that is not compatible with the hypothesis that there is an age-related difference in pacemaker rate (unless such a difference is also accompanied by a recalibration of reference memory). Consider several potential explanations for our age-related differences in duration judgments, which we introduced earlier.
Pacemaker Rate One way to explain the difference between younger adults and older adults is to assume that duration judgments reflect only the rate of a pacemaker or other similar component of an internal clock. Theorists usually assume that a pacemaker that is not rate-compensated should operate more slowly in older adults. A slower pacemaker rate would shorten subjective duration. Our findings are in the opposite direction. Perhaps the pacemaker runs faster in older adults, or perhaps older adults overcompensate for a slower pacemaker. These explanations are clearly post hoc, fairly implausible, and difficult to test. Note that the finding that no significant age-related effects occurred in studies using the reproduction method does not effectively challenge the pacemaker-rate hypothesis. In this method, any individual differences in pacemaker rate present during the target duration are also present during the reproduction, effectively reducing the effect of such individual differences on the duration judgment ratio.
Metabolic Rate or Body Temperature Previous research suggests that experienced duration lengthens as body temperature increases (Hancock, 1993), and it is worthwhile to consider whether our findings can be explained in terms of this effect or allied influences such as basal metabolic rate. Temperature or metabolism may affect a central time-generating mechanism, which may directly affect duration judgments (Hoagland, 1933; Hoagland & Perkins, 1935) or form the basis of a more complex hybrid model in which it is only one component (Treisman, 1963; Treisman, Faulkner, Naish, & Brogan, 1990; Wearden & Penton-Voak, 1995). To account for age-related differences in duration judgments, there must be similar age-related changes in temperature or metabolism. Such changes are in the form of a monotonically decreasing function (Altman & Dittmer, 1968; Kadlub, 1996). However, a temperature or metabolic hypothesis faces the same difficulties as the pacemaker-rate hypothesis: Duration judgments obtained using the methods of verbal estimation and production are the opposite of what is predicted. If temperature and metabolism play a role in duration judgments, they are not the only important influences. At the least, a hybrid model is needed.
Memory Processes Another possibility is that there are age-related differences in memory processes, such as in the rate of forgetting of informa-
tion (cf. Inglis, 1965). In this kind of explanation, the observed differences between participants of different ages reflect a more rapid loss of information concerning events, especially those nearer to the start of a time period. If older adults forget those events at a greater rate than do young adults, they may infer from the decreased memory strength of early events that the duration is relatively longer. Thus, verbal estimates would increase and productions would shorten. One problem with this potential explanation is that the reproduction method should reveal such memory effects, in that a person must remember the target duration at the time he or she is making a reproduction. Indeed, the reproduction method usually reveals large effects in people displaying severe memory disorders (e.g., Richards, 1973). Kelley (1980) and Vanneste and Pouthas (1995, Experiment 2) obtained evidence supporting the notion that memory may play a role in producing age-related differences in the retrospective paradigm: Older adults made shorter reproductions than did younger adults. However, as noted earlier, we were not able to include these data in our meta-analyses. Data included in the present meta-analyses, all of which came from the prospective paradigm, are amenable only to post hoc explanations in terms of age-related memory differences.
Attentional Resources Some evidence suggests that older adults may have more limited attentional resources than do younger adults. Age-related differences in attentional resources are revealed most clearly in divided-attention tasks: Older adults may ordinarily have fewer attentional resources with which to divide attention. Prospective timing requires that a person divide attention between nontemporal (stimulus) information processing and temporal information processing. An attentional-gate model (Zakay & Block, 1997), which includes a pacemaker but emphasizes the role of attention, can explain the present findings as well as the moderating influence of duration judgment method. This model predicts that experienced duration (i.e., duration experience in a prospective paradigm) lengthens as a function of the amount of attentional resources that a person allocates to processing temporal information, as opposed to processing nontemporal (stimulus) information. In this account, verbal estimation and production of duration rely on translations between duration experience and conventional temporal units. These translations are learned in everyday situations. With only one exception, the experiments reviewed here used either no nontemporal information processing task or an easy concurrent task during the target duration. Compared with everyday situations in which older adults must process nontemporal information at a considerable rate, in the present experiments the older adults would have had an unusually large amount of residual resources with which to attend to time. This would increase their verbal estimates and shorten their productions, just as our data show. This explanation predicts that the greater subjective-to-objective duration ratio in older adults should be eliminated or reduced if the time period contains a difficult information-processing task. In fact, McNanamy (1968) found no significant age-related difference in the mean reproduction of 30-s durations spent viewing a series of complex polygons. Polyukhov (1989) reported the typical age-related increase when partici-
BLOCK, ZAKAY, AND HANCOCK
pants performed no task ("empty" interval), but no significant age-related effect when participants performed a relatively difficult verbal task (' 'creating sentences using proposed words"). Craik and Hay (in press) also used a relatively difficult nontemporal information-processing task, and they found the opposite of what is revealed in the present meta-analysis: Older adults gave smaller verbal estimates and made longer productions than did younger adults. Additional studies like these, using tasks varying in difficulty, are needed to help to clarify the role of attentional demands in duration judgments made by younger and older participants. Conclusions The finding that older adults gave a greater duration judgment ratio (larger verbal estimates and shorter productions) than did younger participants is surprising because it is in the opposite direction from what is predicted based on the notion that a biological pacemaker operates more slowly with aging. Existing experimental methodology might have been insufficient to reveal such a biological effect. In any event, an explanation must be sought elsewhere for the oft-repeated idea that time passes more quickly when one ages. James (1890) may have been correct to suggest that different age-related temporal experiences occur over shorter and longer time periods. Perhaps the best explanation for the differences we found here arises from the observation that nearly all experiments used empty durations. The situation in which participants found themselves differed greatly from many everyday situations. If aging is accompanied by a decrease in processing resources, as other evidence suggests, in many everyday situations older adults may ordinarily be sufficiently occupied with information processing that they cannot attend much to time. Because the present experiments tended to lack such attentional demands, the older adults may have found it unusually easy to attend to time during the experimental duration. This would then lengthen their experienced duration because more subjective temporal units would have accrued than is usually the case for a duration having a certain number of objective (verbally labeled) units. Time is central to all behavior; atemporal behavior is an oxymoronic notion. Our findings offer intriguing glimpses into potential explanations of cognitive aging. In addition, they begin to address the largely unresolved problem of individual differences in temporal judgment. The findings reveal that absolute methods of duration judgment (e.g., verbal estimation and production) are distinct from relative methods (e.g., reproduction). Absolute duration judgment methods may clearly reveal individual differences, whereas relative methods are sometimes not sensitive enough to do so. The literature on aging and duration judgment is clearly limited. No published study contained enough data for us to evaluate younger and older adults' retrospective duration judgments; future studies using a retrospective paradigm may help clarify the age-related differences found here in the prospective paradigm. Additional studies (e.g., Craik & Hay, in press) that vary information-processing difficulty, especially at the higher ranges of difficulty, are also critically needed. The present review provides several excellent reasons for researchers to continue to explore age-related differences in duration judgments.
References References marked with an asterisk (*) indicate studies included in at least one meta-analysis. Altman, P. L., & Dittmer, D. A. (1968). Metabolism. Bethesda, MD: Federation of the American Societies for Experimental Biology. Arenberg, D. (19S8). Retention of time judgment in young and old adults. Journal of Gerontology, 23, 35-40. Baum, S. K., Boxley, R. L., & Sokolowski, M. (1984). Time perception and psychological well-being in the elderly. Psychiatric Quarterly, 56, 54-61. *Beck, C. T. (1988). Norm setting for the verbal estimation of a 40second interval by women of childbearing age. Perceptual and Motor Skills, 67, 577-578. Bell, C. R. (1972). Accurate performance of a time-estimation task in relation to sex, age, and personality variables. Perceptual and Motor Skills, 35, 175-178. Block, R. A. (1979). Time and consciousness. In G. Underwood & R. Stevens (Eds.), Aspects of consciousness: Vol. 1. Psychological issues (pp. 179-217). London: Academic Press. Block, R. A. (1985). Contextual coding in memory: Studies of remembered duration. In J. A. Michon & J. L. Jackson (Eds.). Time, mind, and behavior (pp. 169-178). Berlin: Springer-Verlag. Block, R. A. (1989). Experiencing and remembering time: Affordances, context, and cognition. In I. Levin & D. Zakay (Eds.), Time and human cognition: A life-span perspective (pp. 333-363). Amsterdam: North-Holland. Block, R. A. (1990). Models of psychological time. In R. A. Block (Ed.), Cognitive models of psychological time (pp. 1-35). Hillsdale, NJ: Erlbaum. Block, R. A., & Eisler, H. (1998). The compleat bibliography on the psychology of time, 1839-1997 [Machine-readable data file]. Bozeman, MT: Montana State University, Department of Psychology [Producer and Distributor]. Block, R. A., & Pierce, C. A. (1998). Using primary-level statistics to improve meta-analytic interpretations. Manuscript in preparation.
Block, R. A., & Zakay, D. (1997). Prospective and retrospective duration judgments: A meta-analytic review. Psychonomic Bulletin & Review, 4, 184-197. Brown, S. W. (1997). Attentional resources in timing: Interference effects in concurrent temporal and nontemporal working memory tasks. Perception & Psychophysics, 59, 1118-1140. *Bull, D. W. (1973). Effects of aging on temporal experience (Doctoral dissertation, Purdue University, 1973). Dissertation Abstracts International, 34, 2921B-2922B. Cahoon, R. L. (1969). Physiological arousal and time estimation. Perceptual and Motor Skills, 28, 259-268. Carlson, A. J. (1943, November 5). The apparent time acceleration with age. Science, 98, 407. Cohen, J. (1967). Psychological time in health and disease. Springfield, IL: Charles C Thomas. Craik, F. I. M., & Hay, J. (in press). Aging and judgments of duration: Effects of task complexity and method of estimation. Perception & Psychophysics. Craik, F. I. M., & Jennings, J. M. (1992). Human memory. In F. I. M. Craik & T. A. Salthouse (Eds.), The handbook of aging and cognition (pp. 51-110). Hillsdale, NJ: Erlbaum. Dmitriyev, A. S. (1980). Orientirovka cheloveka vo vremeni (osoznannaia otsenka korotkikh intervalov vremeni) [Human orientation in time (conscious estimation of short time intervals)]. Uspekhi Fiziologicheskikh Nauk, 11, 47-67. *Eisler, A. D., & Eisler, H. (1994). Subjective time scaling: Influence of age, gender, and Type A and Type B behavior. Chronobiologia, 21, 185-200.
AGING AND DURATION JUDGMENTS
•Feifel, H. (1957). Judgment of time in younger and older persons.
review of research literatures [Computer software]. Hillsdale, NJ: Erlbaum.
Journal of Gerontology, 12, 71-74. Fitzpatrick, J. J.. & Donovan, M. J. (1978). Temporal experience and motor behavior among the aging. Research in Nursing and Health, I, 60-68.
Joubert, C. E. (1983). Subjective acceleration of time: Death anxiety and sex differences. Perceptual and Motor Skills, 57, 49-50. Kadlub, E. A. (1996). Time: Language and substance. Perceptual and
Fraisse, P. (1963). The psychology of lime (J. Leith, Trans.). New %rk: Harper & Row. (Original work published 1957)
Motor Skills, S3, 903-913. Kausler, D. H. (1991). Experimental psychology, cognition, and human
Fraisse, P. (1984). Perception and estimation of time. Annual Review of Psychology, 35, 1-36.
aging (2nd ed.). New York: Springer-Verlag. Kelley, N. L. (1980). Accelerated time among the aged: A memory
Frankenhaeuser, M. ( 1959). Estimation of time: An experimental study. Stockholm: Almqvist & Wiksell.
hypothesis (Doctoral dissertation, University of Florida, 1980). Dissertation Abstracts International, 41, 2365B.
Gallant, R., Fidler, T, & Dawson, K. A. ( 1991). Subjective time estimation and age. Perceptual and Motor Skills, 72, 1275-1280. Gilpin, A. R. (1993). A program to test equality of two or more coefficients of variation. Behavior Research Methods, Instruments and Computers, 25, 65-66.
Kline, D. W., & Burdick, D. C. (1980). Metronome pacing and age: The effect of irrelevant stimuli. Experimental Aging Research, 6, 393-
398. *Kline, D. W., Holleran, L., & Orme-Rogers, C. (1980). The estimation of short time intervals as a function of age and metronome pacing.
•Goldstone, S., Boardman, W. K., & Lhamon, W. T. (1958). Kinesthetic cues in the development of time concepts. Journal of Genetic Psychology, 93, 185-190.
International Journal of Aging and Human Development, 11, 127-
134. Kuhlen, R. G., & Monge, R. H. (1968). Correlates of estimated rate of
Goldstone, S., Lhamon, W. X, & Boardman, W. K. (1957). The time sense: Anchor effects and apparent duration. Journal of Psychology, 44, 145-153.
time passage in the adult years. Journal of Gerontology, 23,
433. *Landaeta, F. S., Saavedra, M. A., & Simicic, V. Y. (1981). La percep-
Grossman, J. S., & Hallenbeck, C. E. (1965). Importance of time and its subjective speed. Perceptual and Motor Skills, 20, 1161 -1166.
cion del tiempo y su relacion con la edad [Time perception and its
Guyau, J.-M. (1988). The origin of the idea of time. In J. A. Michon,
LeBlanc, A. F. (1966). Orientation and estimation of time as a function
V. Pouthas, & J. L. Jackson (Eds. and Trans.), Guyau and the idea
of age (Doctoral dissertation, Ohio State University, 1966). Disserta-
of time (pp. 93-148). Amsterdam: North-Holland. (Original work published
relation to age]. Revisla Chilena de Psicologia, 4, 35-41.
tion Abstracts International, 27, 1624B-1625B. *LeBlanc, A. F. (1969). Time orientation and time estimation: A func-
Hancock, P. A. (1993). Body temperature influence on time perception. Journal of General Psychology, 120, 197-216.
tion of age. Journal of Genetic Psychology, 115, 187-194. Lejeune, H., & Pouthas, V. (1991). Production de duree chez la personne
Hedges, L. V. (1981). Distribution theory for Glass's estimator of effect
agee: Comparaison avec le jeune adulte [ Response duration differenti-
size and related estimators. Journal of Educational Statistics, 6, 107-
ation in elderly people: A comparison with young adults], Annee
Psycho/ogique, 91, 489-504.
Hedges, L. V. (1982a). Fitting categorical models to effect sizes from a series of experiments. Journal of Educational Statistics, 7, 119-
tual and Motor Skills, 41, 235-238. *Licht, D., Morganti, J. B.. Nehrke, M. F, & Heiman, G. (1985). Media-
Hedges, L. V. (1982b). Fitting continuous models to effect size data. Journal of Educational Statistics, 7, 245-270. Hedges, L. V., & Olkin, I. (1985). Statistical methods for
Lemlich, R. (1975). Subjective acceleration of time with aging. Percep-
tors of estimates of brief time intervals in elderly domiciled males. International Journal of Aging and Human Development, 21, 211-
Orlando, FL: Academic Press. Hicks, R. A., Bramble, W., & Ulseth, S. (1967). Socialization and time perception in aged subjects. Perceptual and Motor Skills, 24, 1170. Hicks, R. E. (1992). Prospective and retrospective judgments of time: A neurobehavioral analysis. In F. Macar, V. Pouthas, & W. J. Friedman (Eds.), Time, action and cognition: Towards bridging the gap (pp. 97-108). Dordrecht, The Netherlands: Kluwer Academic.
225. Light, L.L. (1991). Memory and aging: Four hypotheses in search of data. Annual Review of Psychology, 42, 333-376. Mach, E. (1900). Die Analyse der Empfindungen und das Verhaltniss des Physischen zum Psychischen [The analysis of sensations and the relation of the physical to the mental]. Jena, Germany: Fischer. Manser, M. P., & Hancock, P. A. (1996). Influence of approach angle on estimates of time-to-contact. Ecological Psychology, S, 71-99.
Hoagland, H. (1933). The physiological control of judgments of dura-
MeCormack, P. D. (1982). Temporal coding and study-phase retrieval
tion: Evidence for a chemical clock. Journal of General Psychology,
in young and elderly adults. Bulletin of the Psychonomic Society, 20,
9, 267-287. Hoagland, H., & Perkins, C. T. (1935). Some temperature characteristics in man. Journal of General Physiology, IS, 399-408.
242-244. *McGrath, J. J., & O'Hanlon, J. F., Jr. (1968). Relationships among
Hornstein, A. D., & Rotter, G. S. (1969). Research methodology in tem-
chronological age, intelligence, and rate of subjective time. Perceptual and Motor Skills, 26, 1083-1088.
poral perception. Journal of Experimental Psychology, 79, 561-564.
*McNanamy, E. W. (1968). Effect of novelty on time judgment by
Inglis, J. (1965). Immediate memory, age and brain function. In A. T.
young and aged persons. Journal of Gerontology, 23, 41-44.
Welford & J. E. Birren (Eds.), Behavior, aging and the nervous sys-
Mentzer, C. A., & Schorr, J. A. (1986). Perceived situational control
tem: Biological determinants of speed of behavior and its changes
and perceived duration of time: Expressions of life patterns. Advances
with age (pp. 88-113). Springfield, IL: Charles C Thomas. James, W. (1890). The principles of psychology. New York: Holt. Janet, P. (1877). Une illusion d'optique interne [An illusion of internal
in Nursing Science, 9, 12-20. Michon, J. A., & Jackson, J. L. (1985). Introduction: The psychology of time. In J. A. Michon & J. L. Jackson (Eds.), Time, mind, and behavior (pp. 2-17). Berlin: Springer-Verlag.
perspective]. Revue Philosophique de la France etl'Etranger, 3, 497502.
Morse, C. K. (1993). Does variability increase with age? An archival
Johnson, B. T. (1989). DSTAT: Software for the meta-analytic review of research literatures [Computer software]. Hillsdale, NJ: Erlbaum.
*Neuringer, C., & Levenson, M. (1972). Time perception in suicidal
Johnson, B.T. (1993). DSTAT 1.10: Software fur the meta-analytic
study of cognitive measures. Psychology and Aging, 8, 156-164. individuals. Omega: Journal of Death and Dying, 3, 181-186.
BLOCK, ZAKAY, AND HANCOCK
*Newman, M. A. (1976). Movement tempo and the experience of time. Nursing Research, 25, 273-279. *Newman, M. A. (1982). Time as an index of expanding consciousness with age. Nursing Research, 31, 290-293. Newman, M. A., & Gaudiano, J. K. (1984). Depression as an explanation for decreased subjective time in the elderly. Nursing Research, 33, 137-139. *Nichelli, P., Venneri, A., Molinari. M., Tavani, E, & Grafman, J. (1993). Precision and accuracy of subjective time estimation in different memory disorders. Cognitive Brain Research, I, 87-93. Nitardy, F. W. (1943, July 30). Apparent time acceleration with age of the individual. Science, 98, 110. *Polyukhov, A. M. (1989). Subjective time estimation in relation to age, health, and interhemispheric brain asymmetry. Zeitschrift fur Gerontologie, 22, 79-84. Poppel, E. (1988). Mindworks: Time and conscious experience (T. Artin, Trans.). Boston: Harcourt Brace Jovanovich. (Original work published 1985) Pumpian-Mindlin, E. (1935). Uber die Bestimmung der bewu/ften Zeitschatzung bei Normalen und dementen Epileptikern [On the determination of conscious time estimation in normals and demented epileptics] . Schweizer Archiv fur Neurologie, Neurochirurgie, und Psychiatric, 36, 291-305. Rammsayer, T. H., Lima, S. D., & Vogel, W. H. (1993). Aging and temporal discrimination of brief auditory intervals. Psychological Research, 55, 15-19. Richards, W. (1973). Time reproductions by H. M. Acta Psychologica, 37, 279-282. Salthouse, T. A. (1991). Theoretical perspectives on cognitive aging. Hillsdale, NJ: Erlbaum. Salthouse, T. A. (1994). How many causes are there of aging-related decrements in cognitive functioning? Developmental Review, 14, 413437. Salthouse, T. A. (1996). Constraints on theories of cognitive aging. Psychonomic Bulletin & Review, 3, 287-299. Salthouse, T. A., Rogan, J. D.. & Prill, K. A. (1984). Division of attention: Age differences on a visually presented memory task. Memory & Cognition, 12, 613-620. 'Salthouse, T. A., Wright, R., & Ellis, C. L. (1979). Adult age and the rate of an internal clock. Journal of Gerontology, 34, 53-57. Schroots, J. J. R, & Birren, J. E. (1990). Concepts of time and aging in science. In J. E. Birren & K. W. Schaie (Eds.), Handbook of the psychology of aging (3rd ed., pp. 45-64). San Diego, CA: Academic Press. Shafer, N. J., & Sullivan, J. A. (1986). A simulation study of a test for the equality of coefficients of variation. Communications in Statistics: Simulation and Computation, 15, 681-695. Sharpe, D. (1997). Of apples and oranges, file drawers and garbage: Why validity issues in meta-analysis will not go away. Clinical Psychology Review, 17, 881-901. Smythe, E. J., & Goldstone, S. (1957). The time sense: A normative.
genetic study of the development of time perception. Perceptual and Motor Skills, 7, 49-59. *Surwillo, W. W. (1964). Age and the perception of short intervals of time. Journal of Gerontology, 19, 322-324. Surwillo, W. W. (1968). Timing of behavior in senescence and the role of the central nervous system. In G. A. Talland (Ed.), Human aging and behavior: Recent advances in research and theory (pp. 1 -35). New "itbrk: Academic Press. Tracz, S. M. (1985). The effect of the violation of the assumption of independence when combining correlation coefficients in a meta-analysis (Doctoral dissertation, Southern Illinois University, 1984). Dissertation Abstracts International, 46, 688A. Tracz, S. M., Elmore, P. B., & Pohlmann, J. T. (1992). Correlational meta-analysis: Independent and nonindependent cases. Educational and Psychological Measurement, 52, 879-888. Treisman, M. (1963). Temporal discrimination and the indifference interval: Implications fora model of the "internal clock." Psychological Monographs, 77(13, Whole No. 576), 571-513. Treisman, M., Faulkner, A., Naish, P. L., & Brogan, D. (1990). The internal clock: Evidence for a temporal oscillator underlying time perception with some estimates of its characteristic frequency. Perception, 19, 705-743. Vanneste, S., & Pouthas, V. (1995). Estimation temporelle prospective et retrospective ehez la personne agee: Comparaison avec le jeune adulte [Prospective and retrospective time estimation in elderly people: Comparison with young adults]. Bulletin de Psychologie, 48, 539-543. Walker, J. L. (1977). Time estimation and total subjective time. Perceptual and Motor Skills, 44, 527-532. Wearden, J. H., & Penton-Voak, I. S. (1995). Feeling the heat: Body temperature and the rate of subjective time, revisited. Quarterly Journal of Experimental Psychology: Comparative and Physiological Psychology, 4SB, 129-141. Wearden, J. H., Wearden, A.J., & Rabbitt, P. M. A. (1997). Age and IQ effects on stimulus and response timing. Journal of Experimental Psychology: Human Perception and Performance, 23, 962-979. Whitrow, G. J. (1972). The nature of time. New York: Holt, Rinehart & Winston. Woodrow, H. (1951). Time perception. In S. S. Stevens (Ed.), Handbook of experimental psychology (pp. 1224-1236). New\brk: Wiley. Zakay, D. (1990). The evasive art of subjective time measurement: Some methodological dilemmas. In R. A. Block (Ed.), Cognitive models of psychological time (pp. 59-84). Hillsdale, NJ: Erlbaum. Zakay, D. (1993). Relative and absolute duration judgments under prospective and retrospective paradigms. Perception &. Psychophysics. 54, 656-664. Zakay, D., & Block, R. A. (1997). Temporal cognition. Current Directions in Psychological Science, 6, 12-16. Received November 10, 1997 Revision received March 3, 1998 Accepted April 1, 1998