Serial report and item recognition of novel visual patterns.
Verbal stimuli have been used extensively in the study of immediate serial memory. The familiar bow-shaped serial position curve has been reported using a variety of techniques which include (i) spoken or written serial recall (e.g. Jones, Madden & Miles, 1992; Morton, Crowder & Prussin, 1971; Murdock, 1968); (ii) ordinal recall in which the order of the items must be specified, but output order is not constrained (e.g. Detterman & Brown, 1974); (iii) probed recall using sequential, ordinal or reversed probes (e.g. Avons, Wright & Pammer, 1994; Hitch, 1974; Murdock, 1968; Nairne, Whiteman & Woessner, 1995); (iv) serial reconstruction tasks, in which all the list items are presented at the end of the trial and the participant sorts them into serial order (e.g. Nairne, Riegler & Serra, 1991; Nelson, Reed & McEvoy, 1977) and (v) a verbal counterpart of the serial report task used here, in which an array of items is presented visually and the participant serially indicates the order in which these appeared (e.g. Greene & Crowder, 1988). All these tasks require explicit memory for the serial order in which the list items were presented.
Few, if any, studies of verbal STM have examined pure item memory, so the direct evidence of serial position curves is lacking. However, some indirect evidence is provided by the incidence of item errors in serial recall. Item errors include both omissions (failing to report an item that was in the list) and intrusions (reporting an item that was not presented in the list). Several studies have shown that the probability of an item error in serial memory does not vary across list positions (e.g. Bjork & Healy, 1974; Fuchs, 1969; Healy, 1974; Hitch, 1974). In contrast, studies which have attempted to measure pure order errors, or minimize item load have reported primacy and recency (e.g. Bjork & Healy, 1974; Healy, 1974, 1975; Jones, Farrand, Stuart & Morris, 1995). Typically, these errors show a serial position gradient, in which the probability of reporting an item in a given position decreases with the difference between input and output positions (Fuchs, 1969; Healy, 1974; Jahnke, Davis & Bower, 1989; Lee & Estes, 1977; Murdock, 1968; Nairne et al., 1991, 1995). This is consistent with the theory that serial order errors arise from failures of temporal or order discrimination (e.g. Lee & Estes, 1977).
Visual STM using novel stimuli
Previous investigations of visual memory for novel patterns revealed serial position curves in which there was a recency advantage restricted to the most recent item, and no observable primacy effect (e.g. Broadbent & Broadbent, 1981; Christie & Phillips, 1979; Hines, 1975; Phillips & Christie, 1977a, b). All these studies used conventional recognition memory tests, except Christie & Phillips, who used pattern completion and a drawing task similar to free recall. Phillips & Christie (1977a, b) advocated a duplex interpretation, whereby recency was attributed to a short-term visualization process, whereas the early part of the serial position curve reflected a more stable, long-term component. The absence of primacy is not confined to abstract patterns, but has also been reported with short sequences of natural scenes followed by a two-alternative recognition test (e.g. Weaver & Stanny, 1978).
Walker and his colleagues (Walker, Hitch, Doyle & Porter, 1994; Walker, Hitch & Duroe, 1993) adopted a different technique in which a small set of items were presented repeatedly in a random spatiotemporal sequence, and participants were asked to report the spatial location in which a particular item had been presented. Clear evidence was obtained that the items were encoded visually, and the temporal order serial position curves showed no primacy and one-item recency. Thus although the paradigm differed markedly, the results resembled those of previous item memory tasks. This can be explained by Phillips & Christie's (1997a) duplex theory if the visualization process encodes the spatial location as well as the configuration of the final item.
Phillips & Christie (1977a; see also Phillips, 1983) argued that their atypical serial position curves showed that matrix patterns were not encoded verbally. Broadbent & Broadbent (1981) proposed that the primacy effect observed in serial recall was due to articulation and thus confined to verbal stimuli. Weaver & Stanny (1978) noted that if they had used verbal materials they would have obtained a quite different serial position curve. Walker et al. (1993, 1994) describe their results as typical visual serial position curves. In each case there is the implied assumption that the serial position curve is determined by the nature of the materials. This position ignores the fact that visual memory studies typically employ recognition tests which do not require knowledge of serial order, whereas verbal tests commonly require explicit memory for serial order.
Visual STM using pictures
Several studies have explored serial order in visual memory by using familiar pictures. With this type of material it has been shown that phonological codes support serial order memory (e.g. Cohen, 1972; Manning & Schreier, 1988; Nelson, Brooks & Borden, 1973). The involvement of phonological codes was confirmed in a more extensive study by Nelson et al. (1977), who found that phonological similarity of the labels impaired order recall for both words and pictures, and participants spontaneously adopted the strategy of naming the pictures. They argued that in serial order tasks verbal coding was necessary, since it was well adapted to sequential discrete presentations. However, visual similarity between pictures also impaired performance, especially with fast rates of presentation, suggesting that visual descriptions also played a role.
These visual memory studies have lent support to the view that the serial position curve is the product of the type of materials used: nameable pictures, like words, give bow-shaped serial position curves whereas complex pictures and novel patterns produce flat serial position functions with unitary recency. Again, this ignores the fact that most studies of immediate memory using short series of items have employed serial order tasks in conjunction with verbal materials and item recognition tasks with visual materials. One exception is the study of Waugh & Barr (1989) who studied serial order memory for short lists of scenes. The results showed clear evidence of primacy, plus a large recency effect when the last item in the series was tested first. However, the contribution of naming in this study was unclear. In one experiment the pictures of each series were drawn from the same conceptual category, to discourage naming. This reduced performance compared to the case in which all pictures were from different categories, but the shape of the serial position curve was unchanged. A more unusual study of serial order comparing memory for rapid visual and auditory binary sequences of eight tones or colours was reported by McFarland & Cacace (1995). Their main concern was to determine if recency was enhanced for auditory items, as proposed in some accounts of the modality effect (e.g. Penney, 1989). Although visual sequential memory was poorer than auditory memory when tested by recognition, the serial position curves in the two modalities had similar bowed shapes and decay rates. They suggested that serial position effects are general properties of memory and do not reflect specific sensory mechanisms.
Short-term memory for spatial locations has been studied using variations on the Corsi task. In this task a sequence of familiar or unfamiliar locations is displayed, and the sequence is then recalled by pointing to the locations in turn. Jones et al. (1995) and Smyth & Scholey (1996) both reported bowed serial position curves similar to those obtained in verbal serial recall. Smyth & Scholey (1992) reported only a small effect of concurrent articulatory suppression on spatial span, so it seems unlikely that the serial position curves can be explained by verbal encoding. Jones and his colleagues found that an irrelevant sequence of discrete varying sounds (including non-speech sounds such as tones) disrupted serial recall of both words and spatial locations (Jones et al., 1995). A similar pattern of disruption across modalities was also found for varieties of articulatory suppression performed during the retention interval. A tentative conclusion would be that verbal and spatial serial recall are closely related, possibly by virtue of their serial order demands (Jones, 1993). This striking correspondence across modalities further emphasizes that properties previously ascribed to verbal memory might be more general properties of serial order tasks.
The purpose of the present investigation was to examine serial order memory for novel visual patterns, which have previously been exclusively studied using item memory tasks. To do this, a task was required which was both relatively natural and easy to use, and similar in its demands to verbal and spatial serial recall. The method adopted to test serial order memory was to display all the patterns simultaneously, at randomly assigned locations, and indicate the serial order by pointing to each successive pattern. This technique is referred to throughout as the serial report task.
The first experiment explored visual serial position curves using the serial report task. Because little is known about the limitations on sequential memory for these items, list length was varied between three and six items.
Twenty students of the University of Essex, nine of whom were female, acted as participants and were paid [pounds]2.00 for participating.
The experiment was controlled by a Macintosh Power PC 9500 driving an Iiyama Vision Master 17[inches] Monitor. The program was written in C and all display changes were synchronized to the video raster, which refreshed at 75.6 Hz, using routines described in Pelli & Zhang (1991). Participants made responses using the mouse of the computer. The patterns displayed were 6 x 6 block matrix patterns in which half the cells were filled. Each cell displayed measured 8 x 8 mm. The patterns were generated at random on-line, and were different for each participant. Patterns were drawn as black (filled) cells on an otherwise plain white ground.
The experimental session began with a short practice in which each participant completed six trials with two patterns per series to familiarize them with the procedure. After this, eight trials were presented at series lengths of three, four, five and six patterns per series in ascending order.
Each trial began with the presentation of a fixation point in the center of the screen, for one second. One second after offset of the fixation point, the pattern series was presented. Each pattern was displayed for 117 screen refreshes (1.55 s) followed by a blank interval of 35 screen refreshes (.46 s). At the end of the series all patterns were displayed simultaneously, in locations randomly selected from eight regularly spaced locations around the perimeter of the screen. Locations which were not used remained blank. In the centre of the screen an untilled square the same size as a pattern was used. This served as a 'blank' pattern.
Participants were instructed to learn the patterns and the order in which they were presented. At test they were to use the mouse to indicate the first pattern presented, then the second, etc., until the end of the series, at which point the screen went blank. They were told that if they did not know which pattern came next in the sequence, they should click the mouse on the blank square in the centre of the screen. The data stored by the program consisted of the order in which the patterns were selected at test. The experiment was self-paced. Before each trial, participants were requested to click the mouse in a rectangle drawn in the bottom right corner of the screen. The cursor was then erased, and reappeared during the test at this location, which was not occupied by any test pattern.
At the end of the experiment, participants completed a short informal questionnaire about the task, and then were debriefed about the purpose and nature of the experiment.
(1) Correct responses
The mean number of correct responses at each serial position for each series length is plotted in Fig. 1. It is clear from this graph that performance declined as series length increased, and also that the serial position curves changed as a function of series length. For the shortest series (length 3), performance was close to ceiling and constant across all serial positions. For series lengths of four, five and six items performance showed a typical bow-shaped curve displaying evidence of both primacy and recency.
These findings were confirmed using repeated measures analysis of variance. As series length increased, the mean number of correct serial decisions averaged across all serial positions decreased from 6.93 (86.6 per cent) at series length 3 down to 3.90 (48.75 per cent) at series length 6. This effect of series length on accuracy was very highly significant (F(3,57) = 46.55, MSE = 0.75, p [less than] .001).
For each series length, the number of correct choices (max = 8) was analysed as a function of serial position, and polynomial contrasts were used to examine the shapes of the serial position curves. No serial position effect was found at series length 3 (F(2,38) = 1.96, MSE = .37; p [greater than] .05). For all longer series there was a significant serial position effect (F(3,57) = 9.3, MSE = 1.15, p [less than] .001 at series length 4; F(4,76) = 6.96, MSE = 1.37, p[less than] .001 at series length 5; F(5,95) = 5.99, MSE = 1.90, p [less than] .001 at series length 6). The quadratic contrasts were highly significant at series length 4 (F(1,19) = 17.8, MSE = 1.62, p [less than] .001), at length 5 (F(1,19) = 17.7, MSE = 1.49, p [less than] .001) and at length 6 (F(1,19) = 9.94, MSE = 2.35, p [less than] .01). This confirms the bowed shape of these curves. The linear contrast was non-significant except for the longest series (F(1,19) = 12.35, MSE = 2.31, p [less than] .005), reflecting the decrease in performance across serial positions 1 to 5. The quartic component was significant only for series of length 5, revealing a slight tendency towards a 'W' shape (F(1,19) = 7.3, MSE = .596, p [less than] .05).
(2) Error analysis
Confusion matrices of input and output serial positions were drawn up for list lengths of four, five and six items. Inspection of the matrices showed that errors were often transpositions to nearby serial positions, suggesting a failure of temporal discrimination. The distribution of errors as a function of distance from the correct serial position was examined statistically. The simplest way to do this would be to calculate the probability of making an error at each distance between the stimulus and response serial positions. But this procedure would be biased against large distance values, since these only occur when the input stimulus is at the beginning or end of the list, and the error rate is typically low in these circumstances. Thus the probabilities were made conditional on an error occurring at a particular input position. Conditional probability estimates were obtained for each participant, and were averaged over the cells corresponding to each input-output distance value. The mean values and standard errors are given in Table 1. For all three list lengths examined there is an increased probability of errors being placed in adjacent cells (distance = 1). Repeated measures analysis of variance showed that the effect of distance was significant (F(2,38) = 8.24, MSE = .04342, p [less than] .005 for length 4, F(3,57) = 7.78, MSE = .02184, p [less than] .005 for length 5, and F(4,76) = 6.76, MSE = .0151, p [less than] .005 for length 6). In each case, the pairwise comparison between the probabilities for distance 1 and distance 2 were highly significant (t(19) = 2.81, p = .01, for length 4; t(19) = 4.34, p [less than] .001 for length 5; and t(19) = 4.16, p = .001 for length 6). No other comparisons across successive values of distance approached significance.
Table 1. Experiment 1. Mean conditional probabilities for errors at each distance from correct serial position (standard errors in parentheses) Distance List length 1 2 3 4 5 4 .39 .19 .13 (.05) (.04) (.05) 5 .33 .18 .12 .20 (.02) (.02) (.02) (.05) 6 .28 .15 .16 .14 .09 (.01) (.02) (.02) (.03) (.04)
These results show that bowed serial position curves can be obtained with novel visual patterns when the task requires serial order judgments to be made. There was clear evidence of a primacy effect extending over several items, and also a marked recency effect which was more limited in extent. Performance was much lower than that typically observed in verbal recall or the Corsi task. This is unsurprising because the stimuli were novel patterns, and participants were required to encode a description of each item in addition to specifying its serial position.
In addition the error analysis provides evidence of temporal order confusions. There was a relatively high incidence of confusions between adjacent serial positions, but no compelling evidence of a gradient of confusions across more than one serial position. Hitch (1974) reported a similar confusion function. These results contest the claim that non-verbal materials produce characteristic serial position curves lacking in primacy effects and demonstrating recency only if followed by an immediate test. One possible objection is that the serial order requirement may encourage verbal recoding, so that the serial position curves reflect serial recall of verbal tokens. When completing the questionnaire at the end of the experiment, about half the participants reported trying to name or describe verbally parts of the patterns, although it is by no means clear how much this contributed to performance. The contribution of verbal strategies is addressed in Expt 2, which examines the effect of concurrent articulation on serial report for visual patterns.
The purpose of Expt 2 was firstly, to confirm the results obtained in Expt 1, and secondly to investigate the effect of concurrent articulatory suppression on serial report of visual patterns. When verbal items (letters or words) are presented visually, steady-state articulatory suppression (repeating a single word such as 'the, the, the...') during presentation has the effect of reducing memory span, and removing the sensitivity of serial recall to the effects of phonological similarity (e.g. Murray, 1968) and word length (Baddeley, Thomson & Buchanan, 1975; Coltheart, Avons & Trollope, 1990). A widely accepted interpretation is that suppression prevents visual items gaining access to a phonological store (see Baddeley, 1986, for a review). Since serial order memory encourages phonological recoding, it is possible that memory for pattern serial order in Expt 1 was mediated by phonological codes. Although it is difficult to describe a matrix pattern completely using a verbal description, the present task requires only that the description is adequate to distinguish items in the series. Since the patterns used in these experiments were generated at random, it is possible that a short verbal description could serve this purpose. If so, then rehearsing these descriptions would be a useful way of augmenting serial order memory for the patterns.
In this experiment, participants performed the pattern serial report both under silent conditions, and while repeating the nonsense syllable blah while the patterns were being presented. If participants make use of verbal descriptions of the patterns to remember the order when performing the task in silence, then performance should be impaired by concurrent suppression.
Twenty students of the University of Essex were paid for participating. None of these had participated in Expt 1, 9 were male, and 11 female.
The materials are as described for Expt 1, except that lists of three items were not used.
The basic experimental procedure and the displays used in the presentation and report of the patterns were the same as for Expt 1. The experimental session began with a short practice in which each participant completed six trials with three patterns per series to familiarize them with the procedure. After this, three blocks of eight trials were presented with increasing series lengths of four, five and six items. This entire procedure was presented twice: once under silent conditions and once where the participant engaged in articulatory suppression by rapidly repeating the nonsense syllable 'blah' at least three times per second, while the patterns were presented. Participants were given strict instructions to start suppressing before the first pattern was presented and to continue until after the last pattern had been presented and the test patterns were displayed, If participants failed to comply they were reminded to do this by the experimenter. The order in which the silent and suppression conditions were run was counterbalanced across participants. Otherwise the procedure was as for Expt 1.
(1) Correct responses
The mean number of serial positions reported correctly for each series length is plotted in Fig. 2a (silent condition) and Fig. 2b (suppression condition). The figures indicate clearly that accuracy declined as list length increased, and that all the serial position curves showed a typical bowed shape, similar to those observed in Expt 1. Another prominent feature is that performance is typically lower in the suppression than in the silent conditions.
At each list length a three-way analysis of variance was conducted with order of conditions (suppression first or silent first) as the between-subjects variable, suppression and serial position as the within-subject variables. A significant effect of suppression was found at all list lengths (F(1,18) = 15.38, MSE = 4.48, p [less than] .01, F(1,18) = 23.13, MSE = 4.8, p [less than] .001, and F(1,18) = 7.98, MSE = 3.01, p [less than] .05, for lengths of 4, 5 and 6 respectively). The effect of serial position was also significant for list lengths of 4, 5 and 6 (F(3,54) = 4.73, MSE = 1.09, p [less than] .05, F(4,76) = 13.05, MSE = 1.51, p [less than] .05, and F(5,90) = 12.69, MSE = 1.41, p [less than] .001). In the case of list lengths 4 and 5 there was a significant interaction between order and suppression, which most probably indicates learning during the experiment. This interaction was absent for list length 6 which was the last list length to be tested. No other effects or interactions were significant, most notably there was no interaction between suppression and serial position, indicating that the serial position curves were the same shape for silent and suppression conditions.
Table 2. Experiment 2. Mean conditional probabilities for errors at each distance from correct serial position (standard errors in parentheses) Distance List length Condition 1 2 3 4 5 4 silent .32 .22 .13 (.04) (.04) (.05) 5 silent .31 .16 .23 .14 (.02) (.03) (.04) (.05) 6 silent .28 .16 .14 .16 .12 (.02) (.02) (.03) (.03) (.03) 4 suppr .41 .26 .17 (.02) (.03) (.04) 5 suppr .34 .21 .16 .14 (.02) (.02) (.02) (.04) 6 suppr .29 .16 .16 .12 .12 (.02) (.02) (.02) (.02) (.03) Key. suppr = articulatory suppression.
The shapes of the serial position curves were further analysed using polynomial contrasts. The quadratic contrasts were highly significant at series length 4 (F(1,18) = 7.27, MSE = 1.74, p [less than] .001), at length 5 (F(1,18) = 41.03, MSE = 1.33, p [less than] .001) and at length 6 (F(1,18) = 23.28, MSE = 2.32, p [less than] .001). This confirms the bowed shape of these curves. The linear contrasts were non-significant at list length 4 (F(1,18) = 2.23, MSE = 1.03, p [greater than] .05) but significant at list length 5 (F(1,18) = 12.25, MSE = 1.84, p [less than] .005) and list length 6 (F(1,18) = 14.9, MSE = 1.77, p = .001), reflecting the downward trend in performance with longer lists. The quartic component was significant only for list length 6, revealing a slight tendency towards a 'W' shape (F(1,18) = 7.6, MSE = .697, p [less than] .05). This component, which was also observed in Expts 1 and 5, may reflect a grouping strategy. It should be noted that this component survived concurrent suppression, and hence is unlikely to be a product of verbal labelling.
(2) Error analysis
For each of the six conditions an error analysis was carried out in the same way as for Expt 1. Table 2 lists the means and standard errors of the conditional probabilities, for errors made at each distance from the correct serial position in each condition. Repeated measures analysis of variance showed a highly significant effect of distance in each condition. In this silent condition, the effect of distance was significant for list length 4 (F(2,38) = 5.24, MSE = .0359, p = .01), list length 5 (F(3,57) = 3.68, MSE = .0313, p [less than] .05) and list length 6 (F(4,76) = 5.34, MSE = .0146, p [less than] .005). In the suppression condition there was also a significant effect of distance (F(2,38) = 9.69, MSE = .0287, p [less than] .001 for length 4, F(3,57) = 9.27, MSE = .0166, p [less than] .001 for length 5, and F(4,76) = 8.38, MSE = .0113, p [less than] .001 for length 6).
The results for the silent condition were very similar to those of Expt 1 with respect to the mean levels of performance, the effect of list length and the shapes of the serial position curves. As before, there was clear evidence of a primacy effect extending over several items, and also a marked recency effect which was more limited in extent. The observed serial position curves are again markedly dissimilar to those previously observed in item recognition tasks.
This experiment also revealed a significant effect of steady-state articulatory suppression on performance. The impairment was greater with list lengths of four and five patterns than with six patterns. Since the list length conditions were run in ascending order, this suggests that interference between the two tasks might be reduced by practice. based on our knowledge of the effects of steady-state suppression, the results suggest that a verbal component is used to support serial report of visual patterns. In the questionnaire responses, most respondents (16/20) reported using a verbal strategy on at least some occasions, and only one respondent reported that suppression did not make the task harder. But there was no clear relation between the extent to which verbal processes were reported and the difficulties caused by suppression.
The effect of suppression could also be explained in terms of general interference resulting from an overload of central capacity. The task of remembering these novel patterns is difficult, and it could be the case that any additional load imposed by a concurrent task would be detrimental. This was examined further in Expt 3, in which performance under silent conditions was compared with articulatory suppression and a tapping task often used as a control in studies of verbal span.
The experiment was carried out using a list length of five patterns. Each participant carried out the serial report task three times under silent, suppression and tapping conditions in a counterbalanced order. There were two minor changes to the test procedure: (i) the blank square was omitted from the test display, and (ii) as each test pattern was selected a frame was drawn round it to inform the participant that it had been chosen. These changes were made after observing that participants rarely exercised the blank option in Expts 1 and 2 and reported that they had difficulty remembering which patterns they had previously selected during the test.
Twelve students at the University of Essex participated in this experiment, in return for a small payment ([pounds]2.00) or course credits. Three participants were male, nine female. One participant was left-handed.
Series of five 6 x 6 matrix patterns were presented as in earlier experiments of this series, with the same display timing. At test, all five patterns were displayed simultaneously and participants indicated the order by clicking on each pattern in turn with the mouse.
All participants first completed a short practice consisting of eight trials at the serial report task. The first four trials had series of four patterns, the last four trials had five patterns in the series, Each trial began with presentation of the word 'ready' displayed in the centre of the screen for one second. At the end of the series, all the patterns were displayed in locations around the perimeter of the screen and participants used the mouse to indicate the order of presentation.
In the main experiment 10 serial report trials with five-pattern series were completed in each of three conditions. In the control condition there was no concurrent activity. In the tapping condition, the participants were required to tap with the first two fingers of their right hand on the edge of the laboratory bench at the rate of about three taps per second while the patterns were presented. The speaking (articulatory suppression) condition required the participant to say 'blah, blah, blah...' repeatedly while the patterns were presented. Practice was given at the concurrent tasks before combining them with the serial report task. The first two trials of each condition were regarded as practice and were discarded, leaving eight trials as data. The three experimental conditions were completed in a counterbalanced order. At the start of each experimental trial the words 'ready', 'tap', or 'blah' were displayed according to the condition, for one second, to act as a fixation guide and to remind the participant what action to take during list presentation.
The mean number of correct responses made at each serial position in each condition is shown in Fig. 3. The typical bowed serial position curves were found in all three conditions. Performance was clearly best in the silent condition, worse in the tapping condition and worst of all in the suppression condition.
These observations were confirmed statistically using a 3 (conditions) x 5 (serial positions) repeated measures analysis of variance. This showed highly significant effects of serial position (F(4,44) = 9.7, MSE = 1.91, p [less than] .001) and condition (F(2,22) = 15.6, MSE = 2.81, p [less than] .001). The interaction was not significant, showing that the effect of interference was roughly constant across serial positions (F(8,88) [less than] 1.0).
Pairwise comparisons between the three conditions showed that performance was impaired in the tapping condition relative to the control condition (t(11) = 2.64, p = .023) and that performance in the suppression condition was impaired relative to tapping (t(11) = 2.82, p = .017).
These results confirm those of Expt 2 in showing that articulatory suppression has a detrimental effect on performance in the serial report task. Moreover, the effect of suppression is greater than that observed with tapping, and hence it is concluded that for this type of pattern, there is a verbal component in remembering the order of the patterns. Once again, removing this component does not seem to alter the shape of the serial position curve. This is consistent with the position that phonological recoding is used to support memory for the patterns in this serial order task. An alternative explanation would be that suppression interferes directly with serial order memory. This seems unlikely in view of the fact that suppression is specific in removing phonological effects from verbal memory span. Moreover, in studies of serial spatial memory, using variants of the Corsi task, several studies reported little or no effect of suppression (e.g. Smyth, Pearson & Pendleton, 1988; Smyth & Scholey, 1992), whereas Jones et al. (1995) claim that only changing-state suppression causes interference.
The results so far suggest that the bowed serial position curves observed in serial report reflect temporal order confusions, a position supported by the error analyses of Expts 1 and 2. A less likely, but still possible explanation is that item information varies across serial positions. The following experiment addresses this issue by testing item information for the patterns in the series.
The first three experiments have shown that when asked to indicate the serial order of a sequence of patterns, bowed serial position curves are found showing evidence of primacy and recency effects. In contrast, Phillips & Christie (1977a, b) reported that serial position curves in memory for matrix patterns are essentially flat with a one-item recency effect. The technique they favoured consisted of reverse serial order recognition testing, in which the last item was tested first, followed by the penultimate item and so on backwards throughout the series. This method gives rise to the one-item recency effect as does probed recognition testing, in which one item from the series is tested at random, and probed recall. However, if instructions are given to remember the first item in the series, then the serial position curves change shape and an advantage appears for the first item as well as the last (Phillips, 1983). The interpretation of these various serial position curves is that the advantage for any item is sustained by active visualization of one of the patterns under strategic control. When the last item is tested first, there is an advantage for maintaining the final item, and the recency effect emerges. However, if the instructions are given to favour the first item in the series, then this will be selectively visualized. Therefore as Phillips (1983) showed, changing the test priorities can change the shape of the serial position curve.
The main purpose of the present experiment was to determine the serial position curves for item recognition when a series of patterns was presented and tested in forward serial order. This procedure is a control to ensure that forwards serial order testing does not influence the availability of item information across serial positions. A secondary purpose was to determine the appropriate level of difficulty of the recognition test. In previous experiments, the matrix patterns used were less complex and presented for shorter display times. Both factors are known to affect item memory (Avons & Phillips, 1980; Phillips & Christie, 1977a, b). In the present experiment level of difficulty was manipulated by setting the difference between target and foil at two, four or six cells.
Twenty participants, 12 males and 8 females, took part in the experiment which lasted one hour. They were paid the sum of [pounds]4.00 for participating. One of these participants had previously participated in Expt 1 and one had participated in Expt 2.
The equipment used to display patterns and collect responses was the same as in Expts 1 and 2. The patterns were randomly generated 6 x 6 matrix patterns with 50 per cent of cells filled. Distractor patterns were created from target patterns by changing the values of two, four or six cells, such that the total number of filled cells remained constant.
The experiment consisted of 60 trials, 20 using target-distractor differences of two, four or six cells. These three conditions were randomized, and thus participants were unaware of the test condition until after the series of five patterns had been presented. The timing and display of the pattern series was identical to that in the previous experiments. A series of two-alternative recognition tests were then presented, testing memory for the patterns in forwards serial order. In each test the target pattern and its derived distractor were displayed side-by-side in the centre of the screen. Responses were made by clicking the mouse on the target pattern. Participants were given instructions to concentrate on each pattern as it was presented and to try to remember it, and to select the pattern that they thought had been presented during the series on each subsequent recognition test. Participants were told that the similarity of target and distractor would vary from trial to trial. A rest period of one minute was introduced after 20 and 40 trials, and participants were informed they could take a rest at other times if they wished to do so. The experimental session was preceded by a short practice session of 15 trials which was similar in all respects to the experiment except that the series consisted of four patterns.
The mean number of correct responses at each level of target-distractor difference is shown in Fig. 4. This shows clearly that performance increases as the target-distractor difference increases. However, the serial position curves show no trace of bowing. There is no tendency towards a primacy effect, and if anything, a slight tendency towards a negative recency effect. A two-way repeated measures analysis of variance showed a highly significant effect of task difficulty (F(2,38) = 15.96, MSE = 4.88, p [less than] .001). However, there was no effect of serial position (F(4,76) = 1.32, MSE = 4.25, p [greater than] .05) and no interaction between difficulty and serial position (F(8,152) [less than] 1.0, MSE = 3.56). Thus the analysis confirms the absence of serial position effects, suggesting that item recognition in this experiment does not vary with serial position.
Averaged across serial positions the mean per cent correct performance was 71.7 with a distractor difference of two cells, 76 with a difference of four cells and 81 with a difference of six cells. Since the mean level of performance with four cells is very close to the desired level of 75 per cent, a distractor difference of four cells was used in a direct comparison of serial report and item recognition in Expt 5.
This experiment confirms and extends the studies of item recognition memory for matrix patterns carried out by Phillips and his colleagues. It shows that when tested in forward serial order, the serial position curves are approximately flat. Earlier studies of item recognition showed that performance was constant across all serial positions except the last, under conditions where the final pattern was tested immediately. The present results suggest that any possible advantage for the final item is eroded by the recognition tests for preceding items. This is entirely consistent with the duplex model of item recognition advocated by Phillips & Christie (1977a).
The results so far suggest that serial order report and forwards serial order item recognition give rise to quite different serial position curves when run in separate experiments. Under these conditions participants may be adopting attentional strategies specific to each type of test, and these strategies give rise to different serial position curves. Experiment 5 combined the two kinds of task in a randomized design to eliminate this possibility.
The concern of this experiment is that attention may be distributed differently in the serial report and item recognition tasks. In the case of serial report, performance may be better at the beginning and end of the series because these items received more attention during input. In contrast, the serial position curves for item recognition may be flat because participants either attend to all items, or attend selectively to particular pattern types, irrespective of their serial position. In the present experiment, the two kinds of test were used randomly within a sequence of trials. Participants became aware of the type of test only after the series of items had been presented.
Twenty participants were paid for participating in this experiment, which lasted approximately 45 minutes.
The same apparatus was used as for Expts 1-4. The patterns were 6 x 6 matrix patterns presented as before. Distractors on the item recognition tests differed from the target pattern by four cell values.
The experimental session began with six trials practice at the serial report task with four patterns in the series. The test procedure used was the same as in Expt 3.
A second practice session provided six trials at the item recognition test, again with four items in the series. This was then followed by a third practice in which both types of test were given and there were five patterns presented in the series. The order in which the two types of test occurred was randomized. Ten trials were given in all, five with the item recognition test, and five with the serial report test.
The main experiment consisted of 35 trials, 20 of which were item recognition and 15 were serial report, again in a random order. At the end of the series in the serial report task all five patterns were displayed around the periphery of the screen and participants used the mouse to indicate the order in which they had been presented. As each pattern was selected a frame was drawn round it. In the item recognition task each pattern was presented along with its corresponding distractor and participants were asked to indicate which of these two patterns was presented in the series. Memory for the patterns was tested in forward serial order.
Scores on the item recognition test were based on 20 trials with a chance level of .5, whereas the serial position data were collected from 15 trials with a chance level of .2. The raw scores were linearly transformed to a percentage scale where 0 represents chance performance and 100 is the maximum score. The mean scores for each serial position in the two tests are provided in Fig. 5. This shows a marked difference in the serial position curves for the two tasks. For item recognition, performance is roughly constant for the first four serial positions, and is lower for the last item in the series. In contrast, performance on serial report shows the familiar bowed serial position curve, with a marked positive recency effect.
The transformed scores were analysed by a two-way repeated measures analysis of variance with test type and serial position as the two factors. The results showed no significant effect of test type (F(1,19) = 2.13, MSE = 426.5, p [greater than] .05), showing that the overall level of performance using the transformed scores was very similar. (This similarity reflects the choice of cell difference in the item recognition test, based on performance in Expt 4.) There was also no overall effect of serial position (F(4,76) [less than] 1.0), showing that there is no variation in performance with serial position averaged across the two test types. However, there was a highly significant interaction of serial position with type of test (F(4,76) = 14.27, MSE = 233.0, p [less than] .001), confirming that the serial position curves differed markedly in the two types of test. To reveal more about the shapes of these curves, further analyses were carried out on the item recognition and serial report data separately.
For the item recognition data there was a highly significant effect of serial position (F(4,76) = 4.46, MSE = 339.3, p = .003). Since the data did not violate the sphericity assumption (Mauchly's test, [C.sup.2](9) = 14.1, p = .12), multiple comparisons were made using the studentized range statistic with the pooled mean square error term. Post hoc comparisons using this statistic showed that the final serial position had significantly lower scores than serial positions 1 to 4, which did not differ from each other. Thus the only serial position effect in the item recognition test is the negative recency effect.
For the serial report data, polynomial contrasts revealed a highly significant quadratic component (F(1,19) = 19.46, MSE = 342.83, p [less than] .001) and a weaker quartic component (F(1,19) = 6.14, MSE = 124.1, p [less than] .05). The linear and cubic components both failed to reach significance. These results are quite consistent with those of the previous experiments, although the recency effect appears to be more marked in this case.
Analysis of the errors in the serial report trials confirmed the findings of Expts 1 and 2. When an error was made it was more likely to be misplaced to an adjacent serial position (conditional probability = .32) than to locations two, three or four serial positions away (probabilities = .21, .19, .18 respectively). This finding was again statistically significant (F(3,57) = 8.11, MSE = .011, p [less than] .001).
The results confirm that for visual matrix patterns presented in a temporal sequence, the serial position curves for item recognition and for serial report are quite distinct. A typical bowed serial position curve was found in serial report. In contrast the serial position curve of the item recognition test was flat across the first four serial positions, with poorer performance on the most recent item. This result shows that performance differences between the two kinds of test are not the result of strategic variations operating during presentation. The serial position curves appear to reflect properties of the two kinds of test.
For the serial report condition the serial position curves were similar to those obtained under blocked conditions in Expts 1 and 2, but with superior performance on the final item. However, in the case of item recognition, a marked negative recency effect arose in the present experiment, which was absent in Expt 4. A tentative explanation may be offered using Phillips & Christie's (1977a) duplex theory. In serial report, the final item exists as a STM representation when the test patterns appear. Hence this last item can he matched against the test patterns and the position of the last item noted before identifying the remainder of the patterns in serial order. Several participants spontaneously reported using this strategy. Thus in the serial report task there is no need to memorize the last pattern; a short-term representation is sufficient for this task. In contrast, item recognition uses a strict serial order test, and STM for the final pattern will be overwritten before that item is tested. It is therefore suggested that participants may invest less effort in memorizing the final pattern, a strategy which is detrimental only to the last serial position of the item recognition task, but which enhances correct selection of the final pattern in serial report. If this account is correct then these anomalous end-position effects will be attentuated by making list length unpredictable.
This series of experiments shows that bowed serial position curves arise when visual memory for novel visual patterns is tested by a method which makes explicit demands on serial order information, even when a potential verbal contribution is discouraged by articulatory suppression. In contrast, when no serial order demands are made, and information about the description of novel and unique items is required, then item memory is generally constant across all serial positions except the last. Previous studies have interpreted the relatively flat serial position curves of item recognition tests as characteristic of non-verbal materials. The evidence presented here together with recent studies of serial position curves in serial report of spatial locations (e.g. Jones et al., 1995; Smyth & Scholey, 1996) argues against this explanation.
The task demands of item recognition and serial report differ quite markedly. The item recognition task used here with novel materials places demands only on memory for unique items, which have never been seen before. The decision in this task requires a description of an item in memory, which is precise enough to distinguish it from its (rather similar) distractor. In practice this decision could be made without any consideration of the time at which the item was presented. In contrast, the serial report task places heavy demands on knowing the time or ordinal position at which an item was presented. In this task the limitations on performance arise from imprecise specification of the serial order, giving rise to confusions between adjacent serial positions. This type of confusion is also common for studies of immediate serial recall for verbal materials (e.g. Fuchs, 1969; Jahnke, Davis & Bower, 1989; Lee & Estes, 1977) and also for spatial span (Smyth & Scholey, 1992). Although the serial report task also depends on item memory, the precision has to be sufficient only to distinguish between the unrelated patterns presented on a single trial.
In free recall tasks, which typically use familiar items, the problem is to discriminate items presented on the most recent list from those encountered on previous lists. Thus although explicit serial order is not required, some context must be specified to demarcate the items presented in the current list. The use of temporal context, together with appropriate search strategies may give rise to primacy and recency, which are ubiquitous with verbal lists. Christie & Phillips (1979) reported a free recall experiment using unique matrix patterns. They found serial position curves like those obtained in item recognition, showing no primacy and recency for the final item. However, performance on all but the last item was poor, and the matching of recall attempts to items was somewhat arbitrary. The procedure differed in other ways from standard free recall, and so there are a number of potential explanations for the unusual serial position curves.
Hence the distinctive serial position curves observed in previous studies of visual memory for matrix patterns and other materials do not necessarily represent properties of visual memory. They may simply be a consequence of presenting unfamiliar and unique items, followed by a test for item information. Other studies using visual patterns have presented the same items repeatedly over trials. Under these conditions, additional episodic information must be available at test to indicate that the item was presented on the current trial. Consequently, the serial position curve changes, suggesting that temporal discrimination contributes to performance (Broadbent & Broadbent, 1981; Santiago & Matos, 1994; Wright et al., 1990). The only exception remains the work of Neath (1993; Neath & Knoedler, 1994), in which unique items tested by a yes/no recognition test exhibited serial position effects, sensitive to retention interval and ISI. The reasons for the discrepancy between the present results and those of Neath are unclear and remain to be investigated.
The present experiments have shown that serial position differences between the two tasks still exist when encoding and attentional strategies are the same at input. This does not entail that identical representations are used in the two tasks. Since the impairment of serial report with concurrent articulatory suppression was greater than with a concurrent tapping task, it is possible that phonological descriptions may have contributed to serial report. Previous studies of verbal material suggest strongly that phonological descriptions have privileged status in serial order tasks: (i) performance is impaired when the list items are phonologically similar (e.g. Baddeley, 1986; Conrad, 1964), (ii) the availability of phonological codes encourages temporal order recall (e.g. Altom & Weill, 1977; Freeman, 1975; O'Connor & Hermelin, 1973) and (iii) the requirement for temporal order encourages the use of phonological codes (e.g. Hanson, 1990; Healy, 1974, 1975). Thus one interpretation of the current experiments is that serial order requirements encourage phonological descriptions of the patterns. More generally this would mean that bowed serial position curves observed with visual patterns could be explained in terms of verbal processes, considerably weakening the present argument. However, recent studies at this laboratory show that the serial report task is extremely sensitive to visual similarity of the patterns, and that this similarity effect survives suppression, demonstrating that the serial report task is supported by visual descriptions.
Some accounts of verbal serial recall have suggested that performance is restricted by the decay of item descriptions in STM, offset by the process of rehearsal (e.g. Baddeley, 1986). Item recognition experiments for matrix patterns show that (apart from the one-item recency effect) performance is consistent across serial positions when tested in forward (this paper) or backward (Phillips & Christie, 1977a) serial order. Moreover, there is no clear evidence for iterative rehearsal of the items, such as would sustain decaying traces of the items. Together, these results provide no support for decay of item descriptions, at least over the time course of this task. This apparent absence of decay in item recognition could be taken as strong evidence that item decay does not play a role in the serial report task, and that the serial position curve in the task must be due to order errors. Although this seems likely, some caution should be exercised since these tasks may be supported by different representations.
In general, studies of verbal memory have traditionally associated short-term memory with immediate serial recall (e.g. Baddeley, 1986), whereas visuospatial studies have identified STM with memory for the most recent display (e.g. Frick, 1988). The present study makes clear that this is an arbitrary distinction based on the kinds of task employed, and that the disparate characteristics of visual or verbal STM may arise from this choice of paradigms. If the limitation on serial recall is determined by the limits of temporal resolution, and the limit on recency mirrors the precision with which novel structures or configurations can be specified, then these two measures can be applied to any modality. To take one recent example, Gathercole & Baddeley (1989) proposed a non-word repetition test as an alternative to memory span as a measure of verbal STM. Their rationale was that non-word repetition requires memory for a sequence of phonological units in the same way that memory span requires specification of a series of words. But the account proposed here suggests an alternative: non-word repetition involves recall of the most recent unique item, and hence involves only the creation of a new item description, whereas memory span requires memory for the order of discrete familiar tokens. If this is true then memory span and non-word repetition represent quite distinct operational definitions of short-term memory capacity.
The conclusion of the present study is that the form of the serial position curve for visual patterns depends on the type of task. One wide-ranging implication is that all serial memory tasks share a common process which generates the typical bowed serial position curve. Several well-specified models have been proposed for verbal serial order tasks, based on temporal discrimination or serial order confusions (e.g. Lee & Estes, 1977; Glenberg & Swanson, 1986; Henson, 1996; Henson, Norris, Page & Baddeley, 1996; Neath, 1993). It remains to be seen how readily these and future models will accommodate serial position data in other modalities. The claim is not that all modalities are equivalent; rehearsal opportunities and output processes will vary across modalities and tasks, and these will influence performance. Rather, the central idea is that there may be a common mechanism which constrains serial order memory. The signature of this mechanism is the serial position curve, which has been studied in this paper. Other properties, such as the effect of similarity, complexity and concurrent events may also reveal correspondences across modalities.
This research was supported by the University of Essex Research Promotion Fund. The author is extremely grateful to Sophie Lovejoy, who made many contributions to the project, including running all the experiments. Thanks are due to Dylan Jones, Mary Smyth and the Department of Psychology Memory Club at the University of Essex for many helpful discussions, and to Rik Henson and one anonymous reviewer for comments on the manuscript.
Altom, M. W. & Weill, J. (1977). Young children's use of temporal and spatial order information in short-term memory. Journal of Experimental Child Psychology, 24, 147-163.
Avons, S. E. & Phillips, W. A. (1980). Visualization and memorization as a function of display time and poststimulus processing time. Journal of Experimental Psychology: Human Learning and Memory, 6, 407-420.
Avons, S. E., Wright, K. L. & Pammer, K. (1994). The word-length effect in probed and serial recall. Quarterly Journal of Experimental Psychology, 47A, 207-231.
Baddeley, A.D. (1986). Working Memory. Oxford: Oxford University Press.
Baddeley, A.D., Thomson, N. & Buchanan, M. (1975). Word length and the structure of short-term memory. Journal of Verbal Learning and Verbal Behavior, 14, 575-589.
Bjork, E. L. & Healy, A. F. (1974). Short-term order and item retention. Journal of Verbal Learning and Verbal Behavior, 13, 80-97.
Broadbent, D. E. & Broadbent, M. H. P. (1981). Recency effects in visual memory. Quarterly Journal of Experimental Psychology, 33A, 1-15.
Christie, D. F. M. & Phillips, W. A. (1979). Simple drawing and pattern completion techniques for studying visualization and long-term visual knowledge. Memory and Cognition, 7, 360-367.
Cohen, G. (1972). Serial position effects in the recall of picture sequences. Quarterly Journal of Experimental Psychology, 24, 41-47.
Coltheart, V., Avons, S. E. & Trollope, J. (1990). Articulatory suppression and phonological codes in reading for meaning. Quarterly Journal of Experimental Psychology, 42A, 375-399.
Conrad, R. (1964). Acoustic confusions in immediate memory. British Journal of Psychology, 55, 75-84.
Detterman, D. K. & Brown, J. (1974). Order information in short-term memory. Journal of Experimental Psychology, 103, 740-750.
Freeman, N. H. (1975). Temporal and spatial ordering in recall by five- to eight-year-old children. Child Development, 46, 137-239.
Frick, R. W. (1988). Issues of representation and limited capacity in the visuospatial sketchpad. British Journal of Psychology, 79, 289-308.
Fuchs, A. H. (1969). Recall for order and content of serial word lists in short-term memory. Journal of Experimental Psychology, 82, 14-21.
Gathercole, S. E. & Baddeley, A. D. (1989). Evaluation of the role of phonological STM in the development of vocabulary in children. Journal of Memory and Language, 29, 336-360.
Glenberg, A. M. & Swanson, N. G. (1986). A temporal distinctiveness theory of recency and modality effects. Journal of Experimental Psychology: Learning, Memory and Cognition, 13, 413-425.
Greene, R. L. & Crowder, R. G. (1988). Memory for serial position: Effects of spacing, vocalization, and stimulus suffixes. Journal of Experimental Psychology: Learning, Memory and Cognition, 14, 740-748.
Hanna, A. & Loftus, G. (1993). A modal for the conceptual processing of naturalistic scenes. Canadian Journal of Experimental Psychology, 47, 548-569.
Hanson, V. L. (1990). Recall of order information by deaf signers: Phonetic coding in temporal order recall. Memory and Cognition, 18, 604-610.
Healy, A. F. (1974). Separating item from order information in short-term memory. Journal of Verbal Learning and Verbal Behavior, 13, 644-655.
Healy, A. F. (1975). Coding of temporal-spatial patterns in short-term memory. Journal of Verbal Learning and Verbal Behavior, 14, 481-495.
Henson, R. N. A. (1996). Short-term memory for serial order. Unpublished PhD dissertation, University of Cambridge.
Henson, R. N. A., Norris, D. G., Page, M. P. A. & Baddeley, A. D. (1996). Unchained memory: Error patterns rule out chaining models of immediate serial recall. Quarterly Journal of Experimental Psychology, 49A, 80-115.
Hines, D. (1975). Immediate and delayed recognition of sequentially presented random shapes. Journal of Experimental Psychology: Human Learning and Memory, 1, 634-639.
Hitch, G. J. (1974). Short-term memory for spatial and temporal information. Quarterly Journal of Experimental Psychology, 26, 503-513.
Jahnke, J. C., Davis, S. T. & Bower, R. E. (1989). Position and order information in recognition memory. Journal of Experimental Psychology, Learning, Memory and Cognition, 15, 859-867,
Jones, D. M. (1993). Objects, streams and threads of auditory attention. In A. D. Baddeley & L. Weiskrantz (Eds), Attention: Selection, Awareness and Control. Oxford: Clarendon Press.
Jones, D. M., Farrand, P., Stuart, G. & Morris, N. (1995). Functional equivalence of verbal and spatial information in serial short-term memory. Journal of Experimental Psychology: Learning, Memory and Cognition, 21, 1008-1018.
Jones, D. M., Madden, C. & Miles, C. (1992). Privileged access by irrelevant speech to short-term memory: The effects of changing state. Quarterly Journal of Experimental Psychology, 44A, 645-669.
Lee, C. & Estes, W. K. (1977). Order and position in primary memory for letter strings. Journal of Verbal Learning and Verbal Behavior, 16, 395-418.
McFarland, D. J. & Cacace, A. T. (1995). Comparisons of memory for nonverbal auditory and visual sequential stimuli. Psychological Research, 57, 80-87.
Manning, S. K. & Schreier, H. (1988). Recency and suffix effects in pictures as a function of recall method. American Journal of Psychology, 101, 97-109.
Morton, J., Crowder, R. G. & Prussin, H. A. (1971). Experiments with the stimulus suffix effect. Journal of Experimental Psychology, Monograph, 91, 169-190.
Murdock, B. B. Jr (1968). Serial order effects in short-term memory. Journal of Experimental Psychology, Monograph, 76 (4, Pt 2).
Murray, D. J. (1968). Articulation and acoustic confusability in short-term memory. Journal of Experimental Psychology, 78, 679-684.
Nairne, J. S., Riegler, G. L. & Serra, M. (1991). Dissociative effects of generation on item and order information. Journal of Experimental Psychology: Learning, Memory and Cognition, 17, 702-709.
Nairne, J. S., Whiteman, H. L. & Woessner, L. (1995). Symmetrical cuing effects for item and position information. American Journal of Psychology, 108, 345-358.
Neath, I. (1993). Distinctiveness and serial position effects in recognition. Memory and Cognition, 21, 689-698.
Neath, I. & Knoedler, A. J. (1994). Distinctiveness and serial position effects in recognition and sentence processing. Journal of Memory and Language, 33, 776-795.
Nelson, D. L., Brooks, D. H. & Borden, R. C. (1973). Sequential memory for pictures and the role of the verbal system. Journal of Experimental Psychology, 101, 242-245.
Nelson, D. L., Reed, V. S. & McEvoy, C. L. (1977). Learning to order pictures and words: A model of sensory and semantic encoding. Journal of Experimental Psychology: Human Learning and Memory, 3, 485-497.
O'Connor, N. & Hermelin, B. M. (1973). The spatial or temporal organization of short-term memory. Quarterly Journal of Experimental Psychology, 25, 335-343.
Pelli, D.G. & Zhang, L. (1991). Accurate control of contrast on microcomputer displays. Vision Research, 31, 1337-1350.
Penney, C. G. (1989). Modality effects and the structure of short-term memory. Memory and Cognition, 17, 398-422.
Phillips, W. A. (1983). Short-term visual memory. Philosophical Transactions of the Royal Society of London, B302, 295-309.
Phillips, W. A. & Christie, D. F. M. (1977). Components of visual memory. Quarterly Journal of Experimental Psychology, 29, 117-133.
Phillips, W. A. & Christie, D. F. M. (1977b). Interference with visualization. Quarterly Journal of Experimental Psychology, 29, 637-650.
Santiago, H. C. & Matos, I. (1994). Visual recognition memory in specific learning-disabled children. Journal of the American Optometric Association, 65, 690-700.
Smyth, M. M., Pearson, N. A. & Pendleton, L. R. (1988). Movement and working memory: Patterns and positions in space. Quarterly Journal of Experimental Psychology, 40A, 497-514.
Smyth, M. M. & Scholey, K. A. (1992). Determining spatial memory span: The role of movement time and articulation rate. Quarterly Journal of Experimental Psychology, 45A, 479-501.
Smyth, M. M. & Scholey, K. A. (1996). Serial order in spatial immediate memory. Quarterly Journal of Experimental Psychology, 49A, 159-177.
Walker, P., Hitch, G. J., Doyle, A. & Porter, T. (1994). The development of short-term memory in young children. International Journal of Behavioral Development, 17, 73-89.
Walker, P., Hitch, G. J. & Duroe, S. (1993). The effect of visual similarity on short-term memory for spatial location: Implications for the capacity of visual short-term memory. Acta Psychologica, 83, 203-224.
Waugh, N. C. & Barr, R. A. (1989). Does retention of order require verbal labelling? Experimental Aging Research, 15, 111-121.
Weaver, G. E. & Stanny, C. J. (1978). Short-term retention of pictorial stimuli as assessed by a probe recognition technique. Journal of Experimental Psychology: Human Learning and Memory, 4, 55-65.
Wright, A. A., Cook, R. G., Rivera, J. J., Shyan, M. R., Neiworth, J. J. & Jitsumori, M. (1990). Naming, rehearsal and interstimulus interval effects in memory processing. Journal of Experimental Psychology: Learning, Memory and Cognition, 16, 1043-1059.
|Printer friendly Cite/link Email Feedback|
|Publication:||British Journal of Psychology|
|Date:||May 1, 1998|
|Previous Article:||Believing details known to have been suggested.|
|Next Article:||Repetition priming for face speech images: speech-reading primes face identification.|