Young school-aged children appear to be capable of spontaneously adopting strategies in some situations and require instruction before using strategies in other situations. The present study investigated three interrelated task factors hypothesized to play an important role in whether strategies would be used by young school-aged children. The task factors manipulated were (a) presentation modality of the task (auditory or visual), (b) presence of objects (objects available at presentation or only at recall), and (c) task difficulty (memory loads from 1 to 7 sequences). Seven- and 11-year-old children participated in an external memory task embedded in a story about a "haunted house" and a "friendly ghost" which required memory for spatial relations among objects. Measures of working memory span were also obtained. Results indicated that, overall, the frequency of strategy use for the 7- and 11-year-olds did not differ. However, there was more strategy use in the object available conditions. External strategies without orientation (e.g., pointing or holding) were used more frequently by 7-year-olds than 11-year-olds. However, external strategies with orientation (e.g., arranging) were used more frequently by the 11-year-olds, especially in the visual with objects condition. Measures of working memory however were not related to strategy use. The limitations of the "production deficiency" approach to strategy development and the results for developmental differences in strategy competency of young children were discussed.

A mnemonic strategy is defined as "a set of behaviors specifically initiated to cope with the problem of remembering" (Bray, 1979, p. 716) and is, therefore, an important way of overcoming memory limitations. Numerous studies have shown developmental increases in strategy use (e.g., Bray, Saarnio, Borges, & Hawk, 1994; DeLoache, 1989; DeLoache & Brown, 1983; Fletcher & Bray, 1995). These differences in memory performance may be due, in part, to the acquisition of new strategies and to attempts to generalize strategies to new situations (Siegler, 1991).

Most developmental research on strategy development has found spontaneous strategy use by older school-aged children and little or no strategy use in preschool and young school-aged children (Brown, Bransford, Ferrara, & Campione, 1983). However, younger children can be easily trained to use an appropriate strategy that will improve memory performance. This has led researchers to propose that young children are "production deficient." From this perspective, young children have the competence to produce a strategy but often fail without training (Flavell, 1970). This hypothesis has led to a focus on strategy training without a real understanding of why production deficient children do not use their abilities without training.

This picture of "production deficient" young children (Flavell, 1970; Siegler, 1991) may be incorrect, however, for three interrelated reasons. Siegler (1991) has noted that the question of differences based on the production deficiencies

hypothesis may not be an adequate question. First, this hypothesis can not account for findings from training studies in which subjects trained to use a strategy rarely have the same level of accuracy as subjects who spontaneously use strategies. Secondly, the production deficiency position can not account for the lack of generalization of strategies. Third, other researchers have shown that preschool children and toddlers use strategies without direct instruction (DeLoache & Brown, 1983; DeLoache, Cassidy, & Brown, 1985; Fletcher, 1993; Wellman, Ritter, & Flavell, 1975). For example, in the study conducted by Wellman et al. (1975) young children were told to remember where a toy was hidden under one of several cups. Children displayed behaviors such as looking at and or touching the hiding place. Preschoolers' and toddlers' use of non-verbal external strategies such as pointing or standing near a hidden object's location is a direct attempt to help themselves remember.

Bray et al. (1994) also found spontaneous strategy use in young children and children with mental retardation who would be expected to be production deficient. Seven-year-old and 11-year-old children without mental retardation and 11-year-old children with mild mental retardation listened to a sequence of sentences specifying locations of moveable objects laid out in front of them (e.g., "The eraser is in front of the chair.") At recall, the children were asked to place the objects in their correct locations. All children, including the children with mild mental retardation, used external strategies, such as pointing, holding, and arranging objects to remember the locations of objects and when an external strategy was devised, it was similar for all groups. It was also found that the frequency of external strategy use increased for the 11-year-old children as the number of sentences increased.

With the apparent weakness of the production deficiency hypothesis as a framework for strategy development, four interrelated views of strategy development have emerged that go beyond the production deficiency hypothesis. These positions emphasize the role of various task-related variables and individual differences that either increase or decrease the likelihood that a strategy will be used and move beyond conceptualizing the child as "having" strategies that are not used unless they are trained.

The first view to go beyond the production deficiency hypothesis is that the type of strategy used maps onto the modality of the to-be-remembered information and/or the modality in which the response is to be executed. Baddeley's (1990) theory of working memory provides a basis for expecting differences in the type of strategy used depending on the modality of presentation. Baddeley conceptualizes working memory, the active conscious part of memory, as being a limited capacity system with two subsidiary slave systems, the articulatory loop for verbal information and the visuo-spatial sketchpad for visual/spatial information. Thus, information presented auditorially should be more likely to be encoded in a verbal code by the articulatory loop and, if a strategy is employed, it is likely to be verbally based. In contrast, visually presented information should be encoded into a visuo-spatial code of the visuo-spatial sketchpad and strategy use is likely to be visually/spatially based.

The modality of execution also differs between studies which find production deficiencies and studies which find competencies. A variety of verbally-based tasks (including sequential memory, free recall, memory for paired-associates, sentence memory, and others) have been used in studies showing production deficiencies in preschool and young children. In contrast, a variety of nonverbally-based tasks (including search tasks, executing verbal instructions, hiding objects under cups, and others) have been used in studies showing strategy competencies in preschool and young children. Thus, it is not unreasonable to suspect that there is some characteristic of verbally-based tasks that limits the use of strategies in preschool and young children.

The second view that moves beyond production deficiencies is that strategy use is influenced by the availability of "tools" in the environment. People try to arrange the environment and their activities in ways that organize, minimize, or eliminate searching and reduce processing load (Gauvain, 1993). Tools for arranging the environment, such as maps, are used to reduce search and processing load, and to provide external storage of information. The reduction in processing load that results from the use of such tools may be the motivations for the use of "tools" are effective. Gauvain (1993) contends that material and symbolic tools not only reduce processing load but also transform thinking. The availability of tools provides structure for how an individual attends to and remembers information. For example, what one does when listening to directions for getting to a particular location across town will differ depending on whether or not a pencil and paper are at hand. Paper and pencil are tools that provide support for writing information rather than trying to verbally rehearse it.

It would be expected that situations that provide tools to aid remembering would elicit a different range of strategies as compared to a situation that provides few or no tools to aid remembering. For example, situations providing manipulatives (e.g., miniature objects) should have a broader range of strategy use than situations where manipulatives are not provided. For instance, in a study of external memory strategies in young children, Bray et al. (1994) asked children to remember a series of sentences such as "Put the shoe on the table." The children were provided with manipulatives (small, movable objects) which they could move or arrange while listening to the series of instructions. The manipulatives provided support for numerous strategies (e.g., pointing to, holding the objects, and object arrangement).

It is also expected that when information is presented in a manner that reduces the effort required to execute a strategy (i.e., pictures or maps) there should be an increase in strategy use. Evidence for this line of thinking is provided by studies that have manipulated task variables in order to reduce the processing load involved in executing a strategy. For instance, Ornstein, Medlin, Stone, and Naus (1985) have shown that when children see a single picture containing several to-be-remembered items, children use strategies more frequently and use a greater variety of strategies than when the items are presented sequentially. The sequential presentation evidently places a load on memory not present with simultaneous presentation. Thus, strategy use in children may be inversely related to the extent to which tools are available to reduce processing load. External strategies (i.e., arranging objects in the order in which they are to be remembered), of course, typically reduce the processing load, so when a task enables the use of external strategies, more strategic activity should occur.

The third view to go beyond the production hypothesis is that strategy use is influenced by the degree of task difficulty. With very simple tasks, strategies may not be used but, as task difficulty increases, the likelihood of strategy use should increase. Siegler & Jenkins (1989) have suggested that on easy tasks (e.g., addition problems such as 2+ 1) children try to just retrieve the answer. However, as difficulty is increased, children use "fall-back" strategies (e.g., counting on fingers) in an attempt to reduce processing load. These expectations have been supported empirically (Siegler & Jenkins, 1989). Similarly, Belmont and Mitchell (1987) found that children increased study time in a memory task as task difficulty increased and Bray et al. (1994) found that 11-year-old children were more likely to use an external memory strategy when required to remember four sentences than when required to remember two.

In the fourth view that goes beyond the production deficiency hypothesis, Bray and Turner (1986) suggest that in addition to task variables, there may be individual differences in how children use whatever situational support is available. They suggest that one possible source of individual differences that may be related to strategy use is working memory. Working memory is part of the short-term store that is responsible for maintaining and manipulating information. Consistent with this view, Daneman (1982) suggested that the differentiation between working memory and short-term memory be based on what subjects are required to do with information in memory. A working memory task involves transformation or manipulation of information while maintaining it in memory, (Baddeley, 1990; Case, 1978, 1985; Daneman, 1982), while short-term memory tasks involve no transformation of the information, rather it is held passively in a buffer with a limited number of "slots."

To successfully execute a memory strategy one must have enough capacity to maintain the to-be-remembered information and devise an effective strategy (Guttentag, 1985). Some children, regardless of age, may not use strategies even with contextual support due to the limitations of working memory, but this would be expected to be particularly true of young children. Thus, working memory span may provide an additional individual difference variable, beyond age, that will allow prediction of whether a child will or will not benefit from a strategy and/or use a particular strategy (Case, 1985; Pressley, Cariglia-Bull, Deane, & Schneider, 1987). The preceding discussion of the conceptual background for the present research draws on situational approaches to cognition and is an extension of the position developed by Bray and Turner (1986, 1987). The background suggests that the modality of presentation, the availability of "tools" (e.g., manipulatives), task difficulty, and individual differences in working memory may all influence developmental differences in strategy use. The four potential influences on strategy use have each been examined in separate studies. The purpose of the present study was to examine all four in one study in order to contrast their relative importance.

An extension of the methodology used by Fletcher and Bray (1995) provided a means for investigating the influence of these four variables on strategy use. Fletcher and Bray (1995) told subjects a story about a haunted house and how "magic objects" might help them get rid of "bad ghosts." The children listened to sentences describing an object's location in a room (e.g., "The book is above the candle.") A context for arrangement was created by having sentences with a spatial aspect and having objects available to subjects at all times. The arrangement strategy consisted of placing objects around a central point in the configuration specified in the sentence. Within this context, many subjects devised an arrangement strategy. In the present study, the following variables were manipulated in children 7- and 11-years old: (a) modality of presentation (auditory or visual), (b) availability of manipulatives (present or not present), and (c) processing (memory) load (one to seven objects). Working memory was also measured for each child.

In the present study the following questions were asked: (1) for the three interrelated task factors manipulated (presentation modality, presence of objects [manipulatives], and task difficulty): (a) Will these task factors influence overall frequency of strategy use? (b) Will these task factors influence frequency of specific types of strategies (e.g., external strategies, rehearsal, imagery)? and (2) Will individual differences in working memory be related to strategy use?

Subjects.The participants were 128 7-year-old and 11-year-old children, 64 each from first-grade (M = 6.9; SD = .68) and fifth-grade (M = 10.8; SD = .41) classrooms located in a large metropolitan area of the Southeast. All children were in there age appropriate grade placements. The ethnic makeup of the sample was 87% white and 13% black.

Design. A factorial design of age (7- vs. 11-years), presentation mode (auditory vs. visual presentation), object availability (present vs. absent during presentation), and memory load (1, 2, 3, 5 or 7 objects) was used, with the last factor as a within-subjects variable.

Sixteen subjects from each grade were randomly assigned to one of four between-subjects conditions. There were equal numbers of males and females in each condition. The four conditions for each group were:

Auditory-presentation/objects-available. Subjects were presented sequences of tape-recorded sentences and at the end of the sequence placed the objects in locations specified in the sentence. Subjects have movable objects available while the sentences are presented. See Figure 1.

Auditory-presentation/no-objects. Subjects were presented sequences of tape-recorded sentences but objects were only available at the end of the sequence. See Figure 2.

Visual-presentation/objects-available. Subjects saw a picture, showing where each object was located on a computer screen (representing a room in the old house). After a time equivalent to the auditory presentation of sentences, subjects were asked to place the objects in the locations shown in the picture. The movable objects were available to subjects during the picture presentation. See Figure 3.

Visual-presentation/no-objects. Subjects saw a picture of objects showing where each object was located on the computer screen. The movable objects were only available at the end of a time equivalent to the auditory presentation of sentences. See Figure 4.

In addition each child was administered four working memory measures, two verbal and two visual, each described in the procedure section.

Materials

Working Memory Measures. Four working memory measures were administered. The measures were (a) backwards digit span for verbal working memory span, (b) sentence span for verbal working memory, (c) crooked worm task for visual working memory, (d) Mr. Cucumber for visual working memory. The sequences of digits for backwards digit span and the sentences for sentence memory were presented by a Marantz PDM221 cassette tape recorder. The figure for the crooked worm task was generated by an AST lap-top computer and presented on a 22 cm-diagonal EMC monochrome monitor. Pictures of Mr. Cucumber (21 cm x 31 cm) were mounted on white poster board measuring 26 cm x 36 cm.

Memory Task. A schematic of the apparatus for the memory task is presented in Figure 1. The stimuli will consist of 9 small movable objects (from left to right in Figure 1: coin, jewel, broom, key, candle, apple, fries, shovel, book) placed in a row at the front of a 53 cm x 39 cm x 2 cm piece of white board. In the no-object conditions, subjects saw the same white board as in the objects-available conditions except that the objects sat in a well (45 cm x 9 cm x 2 cm) with a lid (58.5 cm x 12.5 cm) that could be removed at the end of presentation to make the movable objects available. When the lid was on, it was level with the white board. The other end of the white board was attached to a wooden case containing an Amdek monitor with a 35 cm-diagonal screen. The screen, which was used to represent an imaginary room in the old house, was covered with a 5 X 5 matrix of 15 mm-diameter black velcro dots placed in a 19 x 14 cm rectangle with 45 mm between the centers of each dot on the horizontal and 32 mm on the vertical. A white plastic ghost (2 x 3 cm) was attached to the central velcro dot. Surrounding the screen was a blue (right-side) and pink (left-side) border extending approximately 11 cm from the margin of the screen (see Figure 1). A clear plexiglass sliding window (35 x 28 cm) with a yellow poster board border (48 x 40 cm) was moved in front of the monitor to block access to the velcro dot grid yet allow the grid and the blue and pink border to be viewed.

Two Panasonic AG-180 video cameras were used to video-record each session. The first camera (face view), which was attached with a camera arm mounted to the side of the monitor case 63 cm from the center of the participant's chair, provided a view of the child's face. The second camera (side view), placed in a cabinet 137 cm to the participant's right, provided a view of the right side of the child, the white board, the objects on it, and the computer screen. A black cloth, decorated with toy stars, spider webs, and a moon, was loosely draped over both cameras and their supports. In addition to enhancing the "haunted house" effect, the cloth camouflaged the cameras; only the lenses of the cameras were visible. A Marantz PDM221 cassette tape recorder was used to present the sequences of sentences. An Apple IIe microcomputer was used to present the feedback light (48 x 39 mm) behind the appropriate velcro dot.

In the verbal conditions, each to-be-remembered sentence included a noun phrase and a prepositional phrase. The noun phrase consisted of "The" followed by one of the nine movable objects. The prepositional phrase, which followed the noun phrase, always consisted of "is" and one of four prepositions ("above", "below", "on the pink side", and "on the blue side") followed by "the" and a second noun, either the "ghost" or one of the eight remaining movable objects. Sequences such as, "The candle is on the blue side of the ghost," and, "The apple is below the candle" were presented.

In the visual conditions, subjects saw pictures of the movable objects on the computer screen that corresponded to a sequence of the sentences in the auditory conditions. Thus, if subjects in the auditory condition heard the sentence, "The apple is above the ghost," the subjects in the visual condition saw a picture of the apple placed above the ghost. The pictures measured 27.5 x 19.5 cm and were mounted on pieces of pink and blue poster board measuring 49.5 x 48.5 cm representing the pink and blue border. The objects in the pictures were the same size as the movable objects.

Procedure

Session 1. The subjects were tested individually in a mobile laboratory parked on the school grounds. Subjects were administered the four working memory measures. Each child, regardless of condition, received the four working memory measures (2 verbally-based measures and 2 visually/spatially-based measures). The order of working memory measures were counterbalanced. Procedures for each measure were:

Backwards digit span. Subjects listened to a tape-recorded string of digits and, at the end of the string, were asked to repeat the list of digits in the reverse order of presentation (e.g., subjects heard 7, 2, 9 and repeat 9, 2, 7). Sequences started with two digits and increased by one until the subjects missed two sequences with the same number of digits. Working memory span was defined as the maximum number of digits that could be recalled correctly or one less than the number of digits in the terminal trial. Subjects heard one digit every 2 seconds.

Sentence Span. Subjects listened to a series of sentences and were asked to recall the final word in each sentence in the order of presentation (e.g., recall the final word of the first sentence first). Sequences started with two sentences and increased by one until the subjects missed two sequences with the same number of sentences. To help ensure that subjects encoded the entire sentence and were not just focusing on the final word, a question about the content of each sentence was asked after recall of the words. Answers to the questions did not figure into subjects' span. However, if a subject failed to answer one or more questions correctly for each sequence of sentences, the data for this measure were treated as missing. An exception to this rule was made for the first two sequences of sentences. This was due to the fact that subjects were not warned that they would be asked questions about the sentences. During pilot data collection it was discovered that 7-year-olds would become confused and focus on the questions and not final words if told of the questions beforehand. Working memory span was defined as the maximum number of words recalled in the correct order. Consistent with prior research, sentences were presented at a rate of one every 2 seconds (Pressley et al., 1987).

Mr. Cucumber. Subjects were shown a picture of Mr. Cucumber (Case, 1985) with 1 to 7 dots on various body parts for 2 seconds per dot . A Mr. Cucumber with 2 dots would be viewed for 4 seconds. Subjects were then given a Mr. Cucumber without dots and were asked to place dots on the same body parts of Mr. Cucumber (e.g., if shown a dot on Mr. Cucumber's right hand, at recall the subject would place the dot on his right hand). Only pairs of body parts were used. Slight modifications were made from the original Mr. Cucumber developed by Case (1985) to increase the number of body part pairs. An extra set of tentacles and round ends on the corners of the mouth were added. Working memory span was defined as the maximum number of dots placed on the correct body part. See Figure 5.

Crooked worm task. This task was designed to be similar to the tasks described by Lohman (1988) and Salthouse (1992) but easier so younger children would be able to complete the task. Subjects saw a figure (<->) on a computer screen which appeared every second in a different location. At recall subjects were given a sheet of paper and were asked to draw the image they saw created by the figure or more generally connect the dots. The figure appeared at 90 degree angles (from previous location) 50% of the time and at oblique angles 50% of the time.

Working memory span was defined as the maximum number of line segments a subject could remember at the correct orientation from the previous line segment.

Memory Task. After subjects were given the working memory measures, they were introduced to the memory task, which was embedded in the tape-recorded story used by Fletcher and Bray (1995). In this story, the child and an imaginary friend entered an old house, but soon the friend got lost, and a friendly ghost appeared to help find the lost friend. The child was then guided through a "haunted house" by the "friendly ghost" and attempted to remember where "magic objects" were placed in an imaginary room represented by the computer screen. After the child heard the tape-recorded story about the haunted house, the experimenter asked the child to name each of the movable objects. If the child gave an object name other than the one used during presentation (e.g., saying "money" instead of "coin"), the experimenter provided the appropriate name. The child was then given a preposition familiarization task to ensure that each preposition was understood. First the experimenter stated the preposition and turned on the feedback light on the computer screen to illustrate that preposition (e.g., saying "This is above the ghost" while turning on the feedback light behind the velcro dot immediately above the ghost). This was done for all four prepositions. In the auditory presentation condition the experimenter read a sample sentence (e.g., "The coin is below the ghost.") In the visual presentation condition, a picture of the coin placed below the ghost was shown, then the experimenter opened the plexiglass door, allowing the child to place the object on the appropriate velcro dot on the computer screen. The feedback light was then activated. This sequence was repeated for all four prepositions. If the child made any errors during the preposition familiarization task, he/she was corrected by the experimenter and the task was repeated for all four prepositions with new sentences. If after presentation of two sets of four prepositions, a subject was unable to respond correctly on all four prepositions, the experimenter allowed the subject to finish out the first day, but the child did not participate in the second day of testing.

Following the preposition familiarization, each child was told that for each imaginary room, he or she would hear tape-recorded sentences (or see pictures of the objects on the computer screen) and would have to remember where to place the objects in the room (i.e., on the computer screen). Each child was told, "You can do anything that you think will help you to remember where the objects are." In all conditions, at the beginning of each trial, the computer screen was covered by a white poster board. Immediately before each trial, there was a tape-recorded message specifying the number of sentences (e.g., "There are three objects in this room."), followed by a tone to cue the experimenter to remove the white poster board covering the computer screen. In the auditory condition, sentences were presented at the rate of one every 7 seconds and in the visual condition, the pictures of the objects were presented for 7 seconds per object ( e.g., a picture with three objects would be presented for 21 sec). In the auditory condition, the end of a sequence was indicated by a tone 7 seconds after the beginning of the final sentence in the sequence. In the visual condition, this tone occurred at the end of the presentation interval. The experimenter then opened the plexiglass sliding window, and in the no-object conditions, removed the lid covering the movable objects. The child then placed the movable objects on the computer screen representing the room. The experimenter then gave the child feedback by repeating each sentence orally in its order of presentation while simultaneously activating the feedback lights behind the appropriate velcro dots on the computer screen. The subjects were given three practice sequences that increased from one to three sentences. At the end of the session, subjects were told that, now that they had practiced, they could return in a few days to help the "friendly ghost" clean out the "bad ghost."

Session 2. The children returned to the mobile laboratory between 48 to 72 hours after the first session. It was explained that the "friendly ghost" needed the child's help to "clean out the bad ghosts" and that the objects in the house could be used to remove the bad ghosts. The experimenter reviewed the instructions. Each child was again told that he/she could do anything to help himself/herself remember where the objects were. Following the review of the instructions, one practice sequence with two sentences and one with three sentences was presented.

After being given two practice sequences, subjects began the experimental sequences. Subjects heard 1, 2, 3, 5, or 7 sentences or saw a picture with the equivalent number of objects. In 3 blocks of 6 trials, one, two, five, and seven sequences of sentences or objects were presented once during each block with two presentations of three sentences or objects in each block for a total of 18 trials.

Each movable object and preposition was mentioned with approximately equal frequency across sequences. Across children, two different sentence orders were used, each involving different combinations of movable objects, prepositions, and targets.

A brief trial-by-trial interview similar to that used by Bray and Fletcher (submitted) followed each sequence. The initial question, was "What did you do to help yourself remember where the objects go?" If an unambiguous response was given, the next sequence was presented. However, if an ambiguous response was given to the first question, an additional probe was given, "Did you say them to yourself, did you make a picture in your head, or did you do something else?" If the subject responded with "said it to myself" or "made a picture in my head," questioning stopped and the next sequence was presented. If the subject responded with "something else," a third question was asked. The third question was, "What did you do?" After these three questions, the interview for that sequence was stopped and the next sequence was presented. Ericsson and Simon (1984, 1993) have maintained that results of trial-by-trial interviews are valid when (a) the interview is given immediately after the encounter with the relevant information, (b) it is brief, and (c) the relevant material is easily accessible to the individual. Interview data becomes invalid when there are long delays between presentation and interview, the interview is long, and the data are not at the conscious level of the individual. Trial-by-trial interview techniques similar to those used here have been used by McGilly and Siegler (1989) and Bray and Fletcher (submitted) and have yielded results supporting the validity of the children's reports.

Data Reduction

Strategy measures. The video tapes were scored for both external and verbal strategy use and trial-by-trial interview responses during the second experimental session. The child's movements during the period following the presentation of each sentence or picture was scored as (a) no external strategy, (b) an external strategy without orientation (holding, pointing, or manipulating movable objects without orienting them toward the target), or (c) an external strategy with orientation. The latter consisted of holding the objects up to the plexiglass window in front of the screen or arranging them around a central point on the board (or the center of their palm) in the configuration specified by the sequence of sentences or the pictures. The verbal strategies were scored as (a) no observable mouth movement, (b) inaudible mouth movement, (c) an inaudible verbalization (with clear mouthings of object/preposition/target names), or (d) an audible verbalization. The external and verbal strategies were scored separately. Trial-by-trial interview responses were scored as: (a) no report/guessing, (b) just remembered, (c) rehearsal, (d) imagery, (e) pointing, (f) holding, or (g) moving. Two independent judges scored 20% of the subjects' strategies across each sentence of the 18 sequences per session with at least 90% agreement. The strategy measures were the proportion of sentences in a sequence for which a strategy was used.

Accuracy. A correct response was defined as placing the specified movable object on the screen in the location mentioned in the sequence (e.g., for the sequence "The coin is above the ghost", the coin has to be placed on the velcro dot directly above and adjacent to the ghost, not above but to the right or left side of the ghost; the appropriate placement was clarified during the preposition familiarization trials). This same method was used for the picture conditions.

In all analyses, BMDP/386 Dynamic statistical programs were used with = .05. Because preliminary analyses indicated no significant gender effects, gender was omitted from all subsequent analyses. Results for overall frequency of strategy use are presented first followed by observed strategy use, reported strategy use, and accuracy. Relations among strategies and working memory are then presented.

Overall Strategy Use

To assess the question of whether presentation modality, presence of objects, and task difficulty influenced the frequency of strategy use, a measure of overall strategy use was created. If any type of strategy was used at least once on a trial, it received a score of 1. The strategies used to create the overall measure were external strategies with orientation, external strategies without orientation, observed verbalizations, reported rehearsal, and imagery. These strategies covered the whole range of strategies observed and reported. Just remembered and guessing were not included. Also, to increase the number of trials on which each proportion was based, the number of objects in each sequence were grouped into low, medium, or high memory loads: (a) low: sequences with one and two objects; (b) medium: sequences with three objects; and (c) high: sequences with five and seven objects.

Overall frequency of strategy use: Auditory presentation conditions versus visual presentation conditions. A mixed model analysis of variance (ANOVA) was conducted with age (7- and 11-year-olds) and presentation modality (auditory, visual) as the between-subjects variables and memory load (low, medium, and high) as the within-subjects variable. The dependent variable was the proportion of trials for which a strategy was used. Comparisons of the auditory versus visual presentation conditions revealed no difference in the overall frequency of strategy use between the presentation modalities (50% and 54%, respectively). There was a significant age effect, F(1, 121) = 4.93. The 7-year-olds used strategies on 60% of trials while the 11-year-olds used strategies on 44% of trials. A significant main effect for memory load, F(2, 242) = 25.72, revealed that strategy use increased as memory load increased. The means for the low, medium, and high memory loads were 46%, 52%, and 59%, respectively. There was also a significant interaction between task difficulty and age, F(2, 234) = 8.03. Analyses of simple effects revealed that at the low memory load the 7-year-olds used strategies on 57% of trials while the 11-year-olds only used strategies on 35% of trials. A significant difference in the frequency of strategy use was also detected for medium memory loads, with the 7-year-olds again using more strategies than the 11-year-olds, 59% versus 44%. On the high memory loads there was no significant difference between strategy use of the 7-year-olds (63%) and the 11-year-olds (55%). The data reveal that overall younger children are observed and/or report significantly more strategies on the lower memory loads than older children.

Overall frequency of strategy use: Objects available conditions versus no object available conditions. A mixed model analysis of variance (ANOVA) was conducted with age and object availability (objects available, objects not available) as the between-subjects variables and memory load (low, medium, and high) as the within-subjects variable. The dependent variable was the proportion of trials for which a strategy was used. There was a significant main effect of condition, F(1, 121) = 93.34, with an increased level of strategy use in the objects-available versus objects-not-available conditions. Subjects in conditions with objects available used strategies on 77% of the trials, while those in conditions where objects were not available used strategies on 26% of trials. There was a significant age effect again, with 7-year-olds using more strategies than the 11-year-olds, F(1, 121) = 8.08.

A significant effect of memory load was found, F(2, 242) = 25.62, with strategy use increasing as memory load increased. There was also a significant interaction of age and memory load, F(2, 242) = 8.10. Analyses of simple effects showed that strategy use increased more dramatically across memory loads for the 11-year-olds as compared to the 7-year-olds.

It should be noted that in the objects-not-available conditions the opportunity to use an external strategy with orientation was not possible and that external strategy use without orientation was limited to pointing. It was possible, however, that subjects in the objects-not-available conditions would increase alternative strategies such as rehearsal, observed and reported, and imagery to a comparable level of strategy use as in the objects-available conditions. However, there did not seem to be a compensatory effect.

Frequency of Particular Strategies: Observed Strategy Use

For all analyses of observed strategy use, memory load was defined as the number of to-be-remembered objects in a sequence (1, 2, 3, 5, or 7).

External Strategies without Orientation. For this analysis the auditory-with-objects and the visual-with-objects conditions were compared, because these were the only two conditions in which external strategies involving objects could be used. Based on the overall frequency of strategy use analysis, the auditory- and visual-no-objects conditions were not included because the use of external strategies without orientation was limited to pointing in these conditions and thus might artificially create a condition effect. A mixed model analysis of variance (ANOVA) was conducted with age and presentation modality as the between-subjects variables and memory load (1, 2, 3, 5, or 7) as the within-subjects variable. The dependent variable was the proportion of trials for which an external strategy with no orientation was used. There was no significant main effect of condition. There was, however, a main effect of age, F(1, 60) = 22.77, with the 7-year-olds using external strategies without orientation on 61% of trials while the 11-year-olds used external strategies without orientation on only 28% of trials.

Table 1. External Strategy Use Without Orientation for Each Age and for Each Strategy Type.

External Strategies with Orientation. For this analysis only the auditory-with-objects and the visual-with-objects conditions were compared, because the use of external strategies with orientation were precluded in conditions where objects were not available. A mixed model analysis of variance (ANOVA) was conducted with age and presentation modality as the between-subjects variables and memory load (1, 2, 3, 5, or 7) as the within-subjects variable. The dependent variable was the proportion of trials for which an external strategy with orientation was used. There were no significant main effects of condition or age.

A significant main effect of memory load, F(4, 240) = 6.79, was subsumed by a significant three-way interaction of memory load, age, and condition, F(4, 240) = 3.74. In the auditory-with-objects condition, 7-year-olds increased external strategy use as memory load increased, while in the visual-with-objects condition, external strategies with orientation decreased as memory load increased. For the 11-year-olds there was an increase in the use of external strategies with orientation in both auditory-with-objects and visual-with-objects conditions across memory loads. However, the increase was much more dramatic in the visual-with-objects condition, from 16% to 50% versus 8% to 30% for auditory-with-objects condition (see Table 2). For the 11-year-olds there was 20% more external strategy use with orientation on the highest memory load for visual-with-objects condition as compared with auditory-with-objects condition.

Verbal Strategies. All four conditions were included in this analysis. A mixed model analysis of variance (ANOVA) was conducted with age and presentation modality/object availability as the between-subjects variables and memory load (1, 2, 3, 5, or 7) as the within-subjects variable. The dependent variable was the proportion of trials for which a verbal strategy was used. There was a significant main effect of age, F(1, 118) = 6.19. Seven-year-olds were observed using verbal strategies more often than 11-year-olds (20% and 11%, respectively). There was no significant main effect of condition for observed verbal strategy use, however it approached significance, F(3, 118) = 2.41, p<.06. There was a trend for more observed verbalizations in the visual-with-objects condition. The percentage of observed verbalizations for each condition were 11% for auditory-with-objects, 12% for auditory-without-objects, 23% for visual-with-objects, and 17% for visual-without-objects.

Table 2. External Strategy Use With Orientation for Each Condition by Memory Load.

Age	7-year-olds	11-year-olds
Condition
Auditory with Objects
Memory Load
One	13%	8%
Two	24%	20%
Three	22%	27%
Five	27%	25%
Seven	20%	29%
Visual with Objects
Memory Load
One	27%	16%
Two	26%	38%
Three	23%	47%
Five	13%	48%
Seven	14%	50%

Frequency of Particular Strategies: Reported Strategy Use. For all analyses of reported strategy use, memory load was defined as the number of to-be-remembered objects in a sequence (1, 2, 3, 5, or 7).

Verbal Strategies. All four conditions were included in this analysis. A mixed model analysis of variance (ANOVA) was conducted with age and presentation modality/object availability as the between-subjects variables and memory load (1, 2, 3, 5, or 7) as the within-subjects variable. The dependent measure was the proportion of trial for which a verbal strategy was reported. There was a significant main effect of age, F(1, 118) = 8.28, with 11-year-olds reporting more rehearsal than the 7-year-olds, 38% versus 24%. Also a significant main effect of condition, F(3, 118) = 4.77, was revealed, with more reported rehearsal in both no-objects conditions as compared to the with-objects conditions. The means for auditory-without-objects and the visual-without-objects were 38% and 45% compared to the auditory-with-objects and the visual-with-objects means, 26% and 19%.

Imagery. All four conditions were included in this analysis. A mixed model analysis of variance (ANOVA) was conducted with age and presentation modality/object availability as the between-subjects variables and memory load (1, 2, 3, 5, or 7) as the within-subjects variable. The dependent measure was the proportion of trials on which imagery was reported. The only significant main effect was for age, F(1, 118) = 11.85. The 7-year-olds reported using imagery more often than the 11-year-olds, 50% and 31%, respectively.

Accuracy

For this analysis all conditions were included. A mixed model analysis of variance (ANOVA) was conducted with age and presentation modality/object availability as the between-subjects variables and memory load (1, 2, 3, 5, or 7) as the within-subjects variable. The dependent measure was the proportion of sequences for which the correct object was placed in the correct position. There was a significant main effect of age with the 7-year-olds being 60% accurate and the 11-year-olds being 80% accurate, F(1, 117) = 128.88. A significant main effect of condition was also detected, F(3, 117) = 34.87, with the visual-with-objects condition having the highest accuracy. A main effect of memory load was significant, F(4, 468) = 245.88, with accuracy decreasing as memory load increased from 1 to 7.

There were also two significant interactions, a memory load by age interaction, F(4, 468) = 24.56 and a memory load by condition interaction, F(12, 468) = 9.27. For the memory load by age interaction, analyses of simple effects showed that accuracy decreased as memory load increased with the decrease in accuracy more marked for the 7-year-olds as compared to the accuracy of the 11-year-olds. See Table 3.

For the memory load by condition interaction, there was no difference in accuracy between the four conditions on memory loads of one, however as memory load further increased the differences between conditions increased. For all conditions there was a decrease in accuracy as memory load increased from 2 to 7, with the decrease more marked for the auditory conditions as compared to the visual conditions. The most dramatic decrease in accuracy was for the auditory-without-objects condition (see Table 4).

The Relationship Between Working Memory and Strategy Use

Working Memory. Preliminary results indicated that there was no strong correlation or association between verbal measures of working memory and verbal strategy use or between visual/spatial measures of working memory and external or imagery strategy use. Thus, correlations were conducted on object-available versus objects-not-available conditions as the between-subject condition. Memory load was classified as: (a) low, (b) medium, and (c) high. Collapsing across object-availability conditions and memory loads provided a larger N and more power to find relationships.

Table 3. Accuracy for the 7- and 11-year-olds at Each memory load.

Age	7-year-olds	11-year-olds
Memory Load
One	97%	99%
Two	76%	92%
Three	51%	83%
Five	35%	70%
Seven	35%	63%

Table 4. Accuracy for Each Condition at Each Memory Load.

Condition	Auditory with Objects	Auditory no Objects	Visual with Objects	Visual no Objects
Memory Load
One	98%	99%	98%	98%
Two	86%	70%	92%	89%
Three	66%	51%	80%	72%
Five	46%	31%	70%	63%
Seven	43%	24%	71%	59%

There was a positive correlation between age and working memory span, r = .55. However, it was of interest to know if a child's working memory span added any additional information beyond knowing a child's age (i.e., the types of strategies likely to be exhibited). For strategy use, the answer for the 7-year-olds appeared to be yes and for the 11-year-olds the answer appeared to be no. Some additional information is provided by a child's working memory span for 7-year-old children. Overall, though, the most information was gleaned from knowing a child's age.

Closer examination of the 7-year-olds revealed relatively weak support for relations among visual/spatial measures and external strategies. In the objects-available conditions, knowing a child's span on the working memory measure crooked worm correlated positively with moving with orientation at all memory loads (low, medium, and high) and with external with orientation at the medium and high memory loads. However, Mr. Cucumber, the other visual/spatial measure, did not significantly correlate with strategy use at any memory load. In the objects-not-available conditions, crooked worm was positively correlated with both verbal (rehearsal) and external (pointing) strategy use and was consistent at all memory loads. Overall, there was some additional information gained by knowing a child's working memory span in addition to his/her age for 7-year-olds. For the 11-year-olds, however, there was no additional information gained by knowing working memory span in addition to a child's age.

---

---

---

---

---

Discussion

Possibly the most intriguing finding of the present study was the unexpected result that, overall, 7-year-olds were as strategic as 11-year-olds. The overall amount of strategic behavior did not increase with age, but how the particular types of strategic behavior were distributed across strategies and conditions changed with age. Older children used fewer strategies at the lower memory loads but increased strategy use as task difficulty increased. Younger children, however, used strategies more frequently and at similar frequencies across all memory loads. It appears that younger children were less able to modulate their strategy use to fit the demands of the task. Older children, however, were able to modulate their strategy use to fit the demands of the task. This finding is consistent with the results of Fletcher and Bray (1995) who found that overall strategy use of 7- and 9-year-old children without mental retardation and 11- and 17-year-old children with mental retardation was equivalent to that of 11- and 17-year-old children without mental retardation at low and medium memory loads.

The finding of equivalent overall strategy use in 7- and 11-year-old children is important for a number of reasons. First, most studies have been designed to study only one type of strategy (e.g., category clustering) and have found production deficiencies. Studies allowing for multiple strategies, like the task used in the present study, do not find evidence of production deficiencies in young children. Second, this result is consistent with the emerging view that children select among multiple strategies (i.e., Bray et al, 1994; Fletcher & Bray, 1995; Siegler and Jenkins, 1989) rather than using only one strategy at a particular age. Third, equivalent overall frequency of strategy use implies that what change in strategy development are situational processes. This is quite a different view from the "production deficient" child who must be trained to use strategies.

In the present study, although 7- and 11-year-old children had similar overall levels of strategy use, the frequency of particular types of strategies differed. For example, verbal strategies were observed more often in 7-year-olds but were reported more often by the 11-year-olds. This result is consistent with literature on private speech which suggests that with increasing experience individuals are able to internalize their speech and no longer need to say things aloud (e.g., Berk, 1992a).

External strategies without orientation were used more frequently by 7-year-olds than 11-year-olds which is consistent with the results of Bray et al. (1994) and Fletcher and Bray (1995). Interestingly, however, there were no age differences in the use of external strategies with orientation (i.e., arranging). This lack of an age difference in external strategies with orientation is different from previous findings. Both Bray et al. (1994) and Fletcher and Bray (1995) found age-related increases in external strategy use with orientation accompanied by age-related decreases in external strategies without orientation. Both studies used auditory presentation with objects-available while this study varied presentation modality and object availability, but even when the auditory-with-objects condition was examined separately, there were no age differences.

One goal of this study was to determine if the frequency and nature of strategy use would be influenced by four interrelated task factors including: (a) modality of to-be-remembered information, (b) availability of objects, (c) task difficulty, and (d) individual differences in working memory. The factor that had the most influence on the frequency of strategy use was the presence of miniature movable objects during the presentation of the to-be-remembered information. More strategy use was found in conditions where objects were available during the presentation of to-be-remembered information. This result supports Gauvain's contention that the availability of "tools" influences the frequency of strategy use. When objects were available for use during presentation of to-be-remembered information, there was an opportunity to reduce processing load by using the objects to externally represent the information.

Availability of objects also influenced the types of strategies used. To execute external strategies with orientation, objects must be available; thus, their use was limited to conditions where objects were available. However, for external strategies without orientation (such as pointing) objects were not necessary. When objects were made available during the presentation of to-be-remembered information, children used a variety of external strategies, with the 7-year-olds using these more often than 11-year-olds. Again, the more frequent use of external strategies without orientation and the lack of an effect for presentation modality is support for the view that frequency of strategy use and the types of strategies used are influenced by the availability of tools.

Whereas the availability of tools influenced external strategy use, there was evidence that the lack of tools influenced verbal strategy use. Verbal strategies were reported more often in conditions where objects were not available. The lack of movable objects during presentation of to-be-remembered information, as suggested by Gauvain (1993), provided less support for external strategies and increased the likelihood of a verbal strategy being used. However, the frequency of verbal strategy use was less than the frequency of external strategy use in the objects-available conditions. This pattern of results suggests that children do not have a "fixed capacity" for selections among strategies. That is, children do not use strategies with the same frequency regardless of task conditions as if they were selecting from a fixed pool of strategies. Rather, strategies are influenced by a trade-off between the cost and benefits in using a strategy (Siegler, 1991).

Many of the effects for presence of tools were further influenced by task difficulty. For overall frequency of strategy use, strategy use increased as task difficulty (memory load) increased, especially for the 11-year-olds. As predicted by the empirical work of Ornstein and colleagues (1985), when pictures of multiple to-be-remembered objects (which would be expected to reduce processing load) were presented, there was more external strategy use with orientation in the visual-with-objects condition as compared to the auditory-with-objects condition, but this was only true for the 11-year-olds (see Table 2). The 11-year-olds were able to use the additional situational support provided by the visual presentation to increase their use of external strategies with orientation. However, for the 7-year-olds the visual presentation appears to have had the opposite effect. Possibly the younger children overestimated their ability to use other types of strategies to aid their recall, reflected in the 7-year-olds using higher levels of holding and moving without orientation. Thus, the 7-year-olds might have thought they could just hold or move the objects on the high memory loads in the visual presentation and remember. More likely, the "cost" in executing an arrangement strategy increases with the number of objects, so children may have reverted to "fall-back" strategies with less cost (holding and moving the objects without orientation).

Further evidence pointing to the influence of task difficulty and its relationship with object availability is found in patterns of accuracy across each condition. Higher accuracy in the visual conditions would be expected from empirical studies in which pictures of multiple to-be-remembered items reduce the processing load, thereby providing additional support in the form of spatial relationships (Ornstein et al., 1985). Visual presentation eliminates the necessity for the subject to generate a representation of the spatial arrangement. Additional support in the visual-with-objects condition was provided in the form of movable objects, further reducing the processing load by allowing external representations of the to-be-remembered information.

Children in the auditory-without-objects condition had the lowest accuracy. Accuracy was also low in the auditory-with-objects condition but was higher than in the auditory-without-objects condition, apparently because of the additional support for strategy use and reduced processing load provided by the presence of objects. This result suggests that auditory presentation alone provides little support for strategy use and may be especially taxing for younger children. This finding is especially important in light of the fact the majority of studies investigating strategies in young children have used verbally-based tasks (see Bray, Fletcher, and Turner, in press) and, therefore, these studies may have substantially underestimated the strategic ability of young children.

Modality of presentation was not related to overall strategy use but did play a role in the use of external strategies with orientation for the 11-year-old children. Baddeley's theory would suggest that external strategies should be used more frequently when the information is presented in a visual manner and less frequently when information is presented in an auditory manner because external strategies map onto the modality of presentation. The increased use of external strategies with orientation for the 11-year-olds in the visual-with-objects condition as compared to the auditory-with-objects conditions, suggests that Baddeley may be partially correct in his hypothesis. There was little if any support beyond this finding supporting Baddeley's theory that strategy use maps onto the modality of the to-be-remembered information. There were no modality effects for frequency of strategy use for external strategies without orientation (e.g., pointing, holding, or moving), verbal strategies (reported or observed), or imagery. The results suggest that the importance of modality as a factor influencing the nature of strategy use may develop with the experience of remembering information of varying modalities but does not consistently map onto the modality of presentation.

Individual differences in working memory were only weakly related to strategy use. There was no relationship between strategy use and working memory for the 11-year-olds, and there were only a few significant relationships for the 7-year-olds. The results from measures of working memory provide little support for Baddeley's distinction between the visuo-spatial sketchpad and the articulatory loop, visual/spatial and verbal working memories. Only the crooked worm measure consistently correlated with strategy use of the 7-year-olds. This may have included a component of active strategy use. That is, the children with a greater span on this task may have used a visualization strategy that they also used on the external memory task. One other possibility is that the 7-year-olds used more strategies than the 11-year-olds at the low and medium memory loads creating more variability, and this may explain why relations were detected for the 7-year-olds and not for the 11-year-olds.

In sum, the three task factors and individual differences in working memory did play a role in children's strategy use, but to varying degrees. Clearly, the factor that influenced the nature and frequency of strategy use the most was the availability of objects during presentation of to-be-remembered information. These results provide strong evidence in favor of Gauvain's (1993) contention that availability of tools in our environment shapes the way we attend to information and how we try to remember the information. There was also support for Ornstein and colleagues' (1985) position that situations which reduce processing load should show more strategy use. In this study the reduction in processing load was generally provided by the availability of objects, but there was some additional support provided by visual presentation. It seems that when the nature of the to-be-remembered information is spatial in nature, as in the task used in the present study, the presentation of information in a format that highlights the spatial nature (pictures) increases accuracy. Importantly, however, Baddeley's (1990) contention that presentation modality should influence the type of strategy used received only mild support.

Finally, it appears that overall strategy use does not change with age, rather, selection among strategies changes with age and with the nature of the situational support available to the child. These results make it clear that understanding the strategic capabilities of children will require the use of tasks that allow multiple strategies. Tasks that focus on only one type of strategy (especially if it is verbally-based, as in most previous studies) will grossly underestimate the strategy competencies of young children. Children cannot be simply described as production deficient as was previously presumed.

References

Baddeley, A. (1990). Human memory. Boston: Allyn and Bacon.

Belmont, J. M. & Mitchell, D. W. (1987). The general strategies hypothesis as applied to cognitive theory in mental retardation. Intelligence, 11, 91-105.

Berk, L. E. (1992a). Children's private speech: An overview of theory and the status of research. In R. M. Diaz & L. E. Berk (Eds.), Private speech: From social interaction to self-regulation (pp. 17-53). Hillsdale, NJ: Erlbaum.

Bray, N. W. (1979). Strategy production in the retarded. In N. R. Ellis (Ed.), Handbook of mental deficiency: Psychological theory and research (2nd ed., pp. 699-726). Hillsdale, NJ: Erlbaum.

Bray, N. W. & Fletcher, K. L. (submitted). Developmental and intellectual differences in relations between self-report and strategy use. Developmental Psychology.

Bray, N. W., Fletcher, K. L., & Turner, L. A. (in press). Cognitive competencies and strategy use in individuals with mild retardation. In W. E. MacLean Jr. (ed.) Handbook of Mental Deficiency, Psychological Theory and Research (3rd ed.). Hillsdale, NJ: Lawrence Erlbaum.

Bray, N. W., Saarnio, D. A., Borges, L. M., & Hawk, L. W. (1994). Intellectual and developmental differences in external memory strategies. American Journal of Mental Retardation, 99, 19-31.

Bray, N. W. & Turner, L. A. (1986). The rehearsal deficit hypothesis. In N. R. Ellis & N. W. Bray (Eds.), International review of research in mental retardation (Vol. 14, pp. 47-71).

Bray, N. W. & Turner, L. A. (1987). Production anomalies (Not strategic deficiencies) in mentally retarded individuals. Intelligence, 11, 49-60.

Brown, A. L., Bransford, J. D., Ferrara, R. A., & Campione, J. C. (1983). Learning, remembering, and understanding. In P. H. Mussen (Ed.), Handbook on child psychology: Cognitive development (Vol. 3). New York: Wiley.

Case, R. (1978). Intellectual development from birth to adulthood. A neo-Piagetian interpretation. In R. S. Siegler (Ed.), Children's Thinking: What develops? Englewood Cliffs, NJ: Prentice Hall.

Case, R. (1985). Intellectual development: Birth to adulthood. New York: Academic Press.

Daneman, M. (1982). The measurement of reading comprehension: How not to trade construct validity for predictive power. Intelligence, 6, 331-345.

DeLoache, J. S. (1989). Young children's understanding of the correspondence between a scale model and a larger space. Cognitive Development, 4, 121-139.

DeLoache, J. S. & Brown, A. L. (1983). Very young children's memory for the location of objects in a large-scale environment. Child Development, 54, 888-897.

DeLoache, J. S., Cassidy, D. J., & Brown, A. L. (1985). Precursors of mnemonic strategies in very young children's memory. Child Development, 56, 125-137.

Ericsson, K. A. & Simon, H. A. (1984). Protocol Analysis. Cambridge, MA: MIT Press.

Ericsson, K. A. & Simon, H. A. (1993). Protocol Analysis (revised edition). Cambridge, MA: MIT Press.

Flavell, J. H. (1970). Developmental studies of mediated memory. Advances in child development and behavior (Vol. 5, pp. 181-211). New York: Academic Press.

Fletcher, K. L. (1993). External orientation and memory strategy use in preschool children. Unpublished master's thesis, University of Alabama at Birmingham, Birmingham.

Fletcher, K. L. & Bray, N. W. (1995). External and verbal strategies in children with and without mild mental retardation. American Journal of Mental Retardation,99, 363-375.

Gauvain, M. (1993). The development of spatial thinking in everyday activity. Developmental Review, 13, 92-121.

Guttentag, R. E. (1985). The mental effort requirements of cumulative rehearsal: A developmental study. Journal of Experimental Child Psychology, 37, 92-106.

Lohman, D. F. (1988). Spatial abilities as traits, processes, and knowledge. In R. J. Sternberg (Ed.), Advances in the psychology of human intelligence (Vol. 4, pp. 181-248). Hillsdale, NJ: Erlbaum.

McGilly, K. & Siegler, R. S. (1989). How children choose among serial recall strategies. Child Development, 60, 172-182.

Ornstein, P. A., Medlin, R. G., Stone, B. A., & Naus, M. J. (1985). Retrieving for rehearsal: An analysis of active rehearsal in children's memory. Developmental Psychology, 21, 633-641.

Pressley, M., Cariglia-Bull, T., Deane, S., & Schneider, W. (1987). Short-term memory, verbal competence, and age as predictors of imagery instructional effectiveness. Journal of Experimental Child Psychology, 43, 194-211.

Salthouse, T. A. (1992). Mechanisms of age-cognition relations in adulthood. Hillsdale, NJ: Erlbaum.

Siegler, R. S. (1991). Children's thinking. Englewood Cliffs, NJ: Prentice Hall.

Siegler, R. S., & Jenkins, E. (1989). How children discover new strategies. Hillsdale, NJ: Erlbaum.

Wellman, H. M., Ritter, K., & Flavell, J. H. (1975). Deliberate memory behavior in delayed reactions of very young children. Developmental Psychology, 11, 780-787.