Gender differences in the dual-task effects on autobiographical memory retrieval during social...

17
British Journal of Psycholog (1998), 89, 61 1427 0 1998 The British Psychological Society Printed in Great Britain 61 1 Gender differences in the dual-task effects on autobiographical memory retrieval during social problem solving Lorna Goddard*, Barbara Dritschel and Andrew Burton Cognitive Neuropychology Research Unit, Universio of East London, Romford Road, Stratfrd, London El5 4LZ, UK A dual-task paradigm was used to explore the effects of cognitive load on social problem solving and autobiographical memory retrieval. The role that gender may play in mediating the relationship was also examined. Participants performed a secondary task concurrently with two primary tasks: (a) a cueing task, and (b) the Means-End Problem-Solving (MEPS) Task, during which they were required to attend to the memories retrieved during solution generation. Two dual-task conditions were employed in order that two levels of secondary task difficulty could be explored. Level of difficulty proved to be an important factor in the effects of resource reduction on the two primary tasks. Retrieval during the MEPS was effected by both the easy and difficult secondary task whereas retrieval on the cueing task was affected by the difficult task only. The results also showed that females (in contrast to males) favoured a more detailed SPS style using a specific memory database. Consequently, under central executive pressure, females’performance was significantly affected while males’ performance remained largely unchanged. Problem-solving behaviour is widely believed to be a resourceful activity requiring effortful cognitive operations (Sweller, Chandler, Tierney & Cooper, 1990). Accordingly, the detrimental effects of added cognitive load on problem-solving ability have been demonstrated across a wide variety of problem-solving tasks, e.g. syllogistic reasoning (Gilhooly, Logie, Wetherick & Wynn, 1993) and mental addition (Logie, Gilhooly & Wynn, 1994). However, the impact of cognitive load on social problem-solving (SPS) ability has been largely overlooked. One way in which cognitive load may affect SPS is through memory retrieval. Previous research has shown autobiographical memory to play a major role in SPS ability; a failure to retrieve specific memories has been found to relate to poor SPS skill (Evans, Williams, O’Loughlin & Howells, 1992; Goddard, Dritschel & Burton, 1996, 1997). It is widely thought that autobiographical memory retrieval involves a strategic, cyclic process (e.g. Conway, 1992; Kolodner, 1985; Norman & Bobrow, 1979; Schank, 1982; Williams & Hollan, 1981) which is mediated by the central executive component of working memory (Conway, 1992). Specific memory retrieval * Requests for reprints.

Transcript of Gender differences in the dual-task effects on autobiographical memory retrieval during social...

British Journal of Psycholog (1998), 89, 61 1427 0 1998 The British Psychological Society

Printed in Great Britain 61 1

Gender differences in the dual-task effects on autobiographical memory retrieval during

social problem solving

Lorna Goddard*, Barbara Dritschel and Andrew Burton Cognitive Neuropychology Research Unit, Universio of East London, Romford Road, Stratfrd,

London E l 5 4 L Z , UK

A dual-task paradigm was used to explore the effects of cognitive load on social problem solving and autobiographical memory retrieval. The role that gender may play in mediating the relationship was also examined. Participants performed a secondary task concurrently with two primary tasks: (a) a cueing task, and (b ) the Means-End Problem-Solving (MEPS) Task, during which they were required to attend to the memories retrieved during solution generation. Two dual-task conditions were employed in order that two levels of secondary task difficulty could be explored. Level of difficulty proved to be an important factor in the effects of resource reduction on the two primary tasks. Retrieval during the MEPS was effected by both the easy and difficult secondary task whereas retrieval on the cueing task was affected by the difficult task only. The results also showed that females (in contrast to males) favoured a more detailed SPS style using a specific memory database. Consequently, under central executive pressure, females’ performance was significantly affected while males’ performance remained largely unchanged.

Problem-solving behaviour is widely believed to be a resourceful activity requiring effortful cognitive operations (Sweller, Chandler, Tierney & Cooper, 1990). Accordingly, the detrimental effects of added cognitive load on problem-solving ability have been demonstrated across a wide variety of problem-solving tasks, e.g. syllogistic reasoning (Gilhooly, Logie, Wetherick & Wynn, 1993) and mental addition (Logie, Gilhooly & Wynn, 1994). However, the impact of cognitive load on social problem-solving (SPS) ability has been largely overlooked.

One way in which cognitive load may affect SPS is through memory retrieval. Previous research has shown autobiographical memory to play a major role in SPS ability; a failure to retrieve specific memories has been found to relate to poor SPS skill (Evans, Williams, O’Loughlin & Howells, 1992; Goddard, Dritschel & Burton, 1996, 1997). It is widely thought that autobiographical memory retrieval involves a strategic, cyclic process (e.g. Conway, 1992; Kolodner, 1985; Norman & Bobrow, 1979; Schank, 1982; Williams & Hollan, 1981) which is mediated by the central executive component of working memory (Conway, 1992). Specific memory retrieval

* Requests for reprints.

612 L. Goddard, B. Dritschel and A. Btlrton

is thought to be highly dependent on sufficient central executive resource capacity. There is, however, no direct evidence to support this view although research with neurological patients provides a strong case for central executive involvement in autobiographical memory retrieval (e.g. Baddeley & Wilson, 1986 ; Williams & Williams, 1996). Therefore, if central executive resources are reduced and consequently less specific autobiographical memories are retrieved, poorer SPS should result.

The aim of the present research was therefore to investigate the effects of cognitive load on SPS and autobiographical memory retrieval. A further objective was to examine individual differences as a mediator of these effects. One characteristic of social problems which set them apart from traditional problems (e.g. syllogistic reasoning tasks) is their open-ended quality. As stated by D’Zurilla & Goldfried (1971) ‘...most of life’s problems have more than one “correct” solution, that is, they may be handled adequately in a number of different ways.. . ’ There is therefore good reason to expect individual differences in SPS. Although this has been explored extensively as a function of social maladjustment (e.g. Marx, Williams & Claridge, 1992), research investigating individual differences within a normal population is sparse. Likewise, little research has been conducted into individual differences in autobiographical memory. One exception is the work of Sehulster (1988, 1995) which suggests that different memory styles can be identified on the basis of gender. The literature to be reviewed here provides a background within which to investigate gender as a factor in autobiographical memory, SPS and the effects of cognitive load on these processes.

Gender differences in autobiographical memov

From a factor analysis of a 60-item self-report measure of everyday memory ability, Sehulster (1 988) revealed a three-factor solution : verbal memory, autobiographical memory and prospective memory. From these three factors, eight different memory styles were identified, each corresponding to a profile of either high (i.e. has a good memory) or low on each of the factors. Females were more likely to have a memory style which was high on autobiographical memory and low on verbal memory, whereas males had a profile which was low on autobiographical memory but high on verbal memory. Sehulster (1995) found relationships between an individual’s memory style and their responses on a questionnaire designed to assess perceived abilities. The responses from this questionnaire (named Things I Do Well) produced a 12-factor solution after factor analysis. These factors referred to enjoyment/ability in different domains, e.g. spatial orientation, love of music, sensitive to moods of others, fun with numbers and computers, etc. Sensitivity to moods was found to be strongly associated with a memory style that was high in autobiographical memory, and females were more likely to report that they were sensitive to the moods of others than males. This factor has been described as depicting a kind of interpersonal intelligence (Sehulster, 1995). In interpreting his findings, Sehulster draws on the work of Tannen (1990) who argues that in a social setting males are more likely to establish a hierarchy, or pecking order, whereas females are more likely to engender a sense of community by sharing their past experiences. Sehulster suggests that

Gender and social problem solving 613

females access emotional and autobiographical material more in their conversations and therefore it is better rehearsed and organized. In contrast, males more frequently access factual information in their conversations resulting in the greater rehearsal and organization of verbal and factual material.

Developmental research suggests that socialization factors may play a large part in gender differences in autobiographical memory. Parental talk about the past serves to teach the child how to formulate and verbalize memories (Nelson, 1991). Longitudinal studies have shown that individual differences in styles of mother-child memory talk have been found to relate to individual differences in the later memory performance of children. For example, children of elaborative parents (i.e. those who talked about events in a narrative form-in terms of who, what, where, when) contributed more to later memory talk than children of non-elaborative parents (Engel, 1986; Reese, Haden & Fivush, 1993). Moreover, parents have been shown to differ in their style of talking about the past depending on whether they are parents of daughters or sons. Reese & Fivush (1993) found that parents of daughters were in general more elaborative than parents of sons and, in turn, daughters participated in memory talk to a greater degree than sons. Further evidence for gender differences in the socialization of children through memory talk have been shown by Reese (1996). She coded mother-child conversations about the child’s birth and noted that mothers of daughters emphasized the interpersonal aspects, whereas mothers of sons focused more exclusively on the child.

Sehulster’s work, together with developmental research, provides a strong rationale for investigating gender differences in autobiographical memory retrieval style during SPS. As females (relative to males) perceive themselves as sensitive to the moods of others and high on the ability to recall autobiographical material and, as from an early age, they are encouraged to encode the interpersonal aspects of events in a narrative form, there is good reason to suggest that this effect may extend to the use of autobiographical memory in SPS. How may this effect be manifested?

Autobiographical memory is believed to be hierarchically organized. The major dimension within this hierarchy is the specific versus the general. Specific memories lie at the bottom of the hierarchy and represent a highly detailed reconstruction of experience. Typically, they are operationally defined as memories referring to one particular day (e.g. ‘last Boxing Day with my family’). General memories are more abstract. Two types have been identified by Williams & Dritschel (1992): categoric memories which describe multiple occurrences of events (e.g. times I’ve gone to the cinema) and extended memories which describe events extended over a longer time period than one day (e.g. my holiday in Greece). Specific memory retrieval is believed to be more resource demanding than retrieving general memories because specific memories reside further down the hierarchy. Specific memory retrieval first entails the formation of a general memory description (Conway, 1992). This description is then used to identify and verify specific cases in long-term memory.

Specific memories are believed to play a special role in SPS because their rich database provides the necessary cues for (a) defining and understanding present problems and (b) generating alternative solutions to problems (Evans e t a/., 1992; Williams, 1996). Our previous research has supported this standpoint (Goddard e t al., 1996, 1997) ; retrieving specific memories while generating solutions to

614 L. Goddard, B. Dritschel and A. Burton

hypothetical problems (the Means-End Problem-Solving Task) was positively correlated to the quality of solutions offered. As the work of Sehulster suggests that autobiographical material is more readily accessed and better organized in females, relative to males, we therefore predict that they will have faster access to the specific memory database (as measured by a cueing task) and will make more use of the beneficial effects of specific memories during SPS (as measured by subjective reports of retrieval during the MEPS). If gender differences do exist in autobiographical memory retrieval, then the effects of added cognitive load on this process may well have different consequences for males and females. Added cognitive load should reduce specific retrieval and increase general retrieval, since the former is more resource demanding than the latter. If females favour the specific memory database for SPS, relative to males, then cognitive load may have a greater impact on retrieval for females than it does for males.

Gender dzfferences in social problem solving (SPS)

Gender differences in the processing of social information may well bear on gender differences in SPS. Indeed, the research here compliments the work of Sehulster. Lykes (1985) found gender differences in self-concept, with females more likely to perceive themselves as interactive and embedded within a social context, and males more likely to perceive differences between themselves and their social context (Woike, 1992). Other relevant findings come from the depression literature. In attempting to account for the development of gender differences in depression rates, Nolen-Hoeksema & Girgus (1994) suggest that although the number of social challenges increases during adolescence for both males and females, the increase is much larger for females. Furthermore, in a survey (Nolen-Hoeksema & Larson, 1992) it was found that females articulated more concern than males in the majority of different challenge domains (e.g. intimate relationships and achievement). In a related area, research has also found that self-esteem is linked to different processes for males and females with males’ self-esteem linked to personal achievements and females’ arising from connections and attachments to others (Josephs, Markus & Tafarodi, 1992). Overall, gender differences seem apparent across many aspects of social functioning and this effect may well extend to SPS. Specifically, we expect that as females are purported to have more experience of challenges, show greater concern in different problem domains and are more interpersonally oriented, their problem-solving style is likely to be more developed. Our measure of SPS was the MEPS. This requires participants to generate solutions to hypothetical problems. Solutions are marked according to the number of details (i.e. means) they contain and the effectiveness of the strategy. We therefore predict that females will generate more effective solutions which contain more means.

The dual-task paradigm

In order to manipulate cognitive load we employed the dual task paradigm, a method which has proved useful in manipulating attentional resources available for tasks. Participants perform the basic primary task of interest concurrently with a secondary

Gender and social problem solving 615

task. It is assumed that non-automatic tasks utilize cognitive resources and that if two tasks draw on the same cognitive processes they compete for resource allocation from a limited resource capacity (Guttentag, 1989). The reduced resource allocation leads to a decrement in performance quality on either one or both of the tasks.

Autobiographical memory retrieval and problem solving is believed to be dependent on central executive resources (Shallice, 1988). Identifying a secondary task which draws exclusively on central executive processes has proved difficult (Toms, Morris & Ward, 1993). The central executive is the least understood component of working memory (Baddeley & Hitch, 1994) but is largely thought of as synonymous with Norman & Shallice’s (1 980) supervisory attentional system (SAS). Like the SAS, it is conceptualized as a general attentional resource system involved in reasoning, decision making and the allocation of cognitive resources to task demands. Individual tasks which have been identified as involving a prominent central executive component are random generation and card sorting (Grant & Berg, 1948).

It was decided to use a task with two levels of difficulty in order to investigate differential effects of load on the two primary memory tasks, i.e. a cueing task (where the task is to retrieve specific memories to word cues) and retrieval during the MEPS. There is good reason to assume that the retrieval strategies underlying these two tasks may follow different principles. The cueing task requires a deliberate search strategy with a readily constructed cue. However, retrieval during the MEPS is not an explicit task requirement. Reporting on the thoughts and images experienced while problem solving more closely fits Bekerian & Dritschel’s (1992) criteria of an indirect, unintentional test of memory. They suggest that retrieval under these more naturalistic conditions may not operate under the same processes as those governing tasks such as the cueing task. Subsequently, these two tasks may be differentially affected by added cognitive load. In particular, we hypothesize that a more demanding secondary task may be necessary for producing effects on the cueing task than during the MEPS. This is because, in the former, the participant is readily provided with the description used to search autobiographical memory whereas, in the latter, the use of autobiographical memory and search strategy must be initiated by the participant.

As research into autobiographical memory using a dual-task paradigm is in its infancy, deciding on an appropriate secondary task is complex. Since the primary tasks of the present study involved overtly articulating autobiographical memories and/or problem solutions, random generation would interfere with articulation. The Wisconsin Card Sorting Task would also present problems in implementation. Instructing participants on their card-sorting performance while orally presenting problem vignettes would be difficult. Our main goal was to reduce attentional resources and interfere with the mechanism purported to be involved in specific retrieval (i.e. the central executive). We required a task which would cause a general, rather than a specific, interference. This is because a task selectively interfering with retrieval from long-term memory (e.g. a semantic verification task) may serve to cue rather than block autobiographical memories. We also required a task which required a continuous response so that participants would not be able to develop a strategy whereby they could switch resources between the primary and secondary tasks.

61 6 L. Goddard, B. Dritscbel and A. Burton

The secondary tasks chosen were computerized, four-choice reaction time tasks. The easier task was designed to incorporate key elements of a simple card-sorting task (i.e. without category change) used by Baddeley, Lewis, Eldridge & Thompson (1 984) in their series of experiments investigating dual-task effects on encoding and retrieval. Baddeley e t a/. found no effects of this task on retrieval accuracy (but an effect on latency). However, it has been argued that this was largely due to the memory tests used, which relied less on strategic retrieval processes and more on associative/cue-dependent processes (Moscovitch, 1994).' More recently, a computerized four-choice reaction time task has also been used by Craik, Govoni, Naveh-Benjamin & Anderson (1996) in order to measure the attentional costs of encoding and retrieving. Craik e t al. also found that retrieval was little affected by dual-task costs and argued that some types of retrieval from episodic memory are automatic. Specifically their findings suggest that recognition memory is automatic, while strategic processes are involved to some extent in cued recall and to a greater extent in free recall. Since autobiographical memory retrieval is widely believed to be a strategic process, it should be sensitive to resource costs and therefore the above null findings are of little concern here. The harder secondary task was a modification of the easy task. Difficulty was increased by (a) invoking a change in response strategy, (b) providing immediate feedback on errors and omissions and (c) changing the rate of stimulus presentation.

In summary, the main purpose of the study was to examine the effects of reduced central executive capacity on memory retrieval (as measured by the cueing task), SPS (as measured by the MEPS) and memory during SPS (as assessed by participants' self-reports). It is predicted that performing the cueing task concurrently with a secondary task will (a) increase response latencies, (6) reduce the number of correct (i.e. specific) responses and (c) increase the number of incorrect (i.e. categoric and/or extended responses). In terms of SPS and autobiographical memory retrieval, it is predicted that the secondary task will (a) have a detrimental effect on the quality of solutions, (b) reduce the number of specific memories retrieved and (c) increase the number of categoric and/or extended memories retrieved. In addition, we examined gender differences and hypothesized that females will outperform males in the absence of added cognitive load but consequently cognitive load will have a greater effect on problem solving and memory for females relative to males. Finally, we hypothesized that retrieval on the cueing task and MEPS will be differentially affected by cognitive load with the cueing task requiring a greater reduction in resources for retrieval to be affected.

Method

Design A repeated measures factorial design was used with one factor (between-subjects) of task condition consisting of three levels (single, easy dual, difficult dual) and a second factor of gender (ix. males, females).

Baddeley e t al. (1984) argue that retrieval which is relatively automatic may be less affected by concurrent load and less dependent on central executive control.

Gender and social problem solving 617

Participants Sixty students (30 females, 30 males) from the University of East London were randomly assigned to one of the three groups, each of which comprised an equal ratio of females to males. Mean ages did not differ between the groups (single task: M = 25.15, SD = 5.32; easy dual task: M = 26.3, SD = 7.36; difficult dual task: M = 27.3, SD = 7.61).

Materials and procedure Autobiographical Memoy Cueing Task. This comprised 10 words (five positive : happy, safe, interested, successful, surprised; five negative: sorry, angry, clumsy, hurt, lonely) which were verbally presented to the participant. Words were presented in the order as listed above with positive and negative words alternating. For each word participants were required to retrieve a specific memory as quickly as possible. A specific memory was described to the participant as a memory for a unique event referring to one particular day. Examples of appropriate specific memories and inappropriate general memories were given. When participants responded with a general memory, they were prompted to recall one particular time. Latencies to the first word of each response were recorded using a stopwatch. Where a memory was inappropriately general and a prompt for a specific memory was given, latencies to subsequent responses to cues were accumulated. A maximum latency of 60 seconds was allowed for each word. If a participant had not recalled within this time, a score of 60 was given and the response was marked as a ‘failure’. Memories were categorized according to the criteria below.

Memoy categorization. 1. Specific memories : A memory was rated specific if the participant recalled an individual event. The criterion of an experience which lasts less than a day has frequently been used to define a specific memory (e.g. Williams, 1992) and was therefore used in this study. 2. Extended general memories : A memory was classified as extended if it referred to a specific event which lasted more than one day. A specific event in this categorization was conceived as something which had a definite beginning and a definite end. 3. Categoric general memories : A memory was rated categoric if it referred to a series of repeated events, e.g. ‘playing squash on Friday nights’ (Williams, 1996).

Memories were categorized by the first experimenter. Our previous research (Goddard e t al., 1996) has found very high inter-rater reliability, i.e. Cohen’s kappa = .98.

Means-End Problem Solving (MEPS) . This consists of 10 vignettes which first describe a problem situation and then the resolution of that problem situation. Problems revolve around social themes. Participants are required to describe the ideal strategy in order to solve the problem and reach the goal. Platt & Spivack (1975b) have shown that it is not necessary to administer all 10 vignettes and many studies have used a shortened form. This study used the same four situations as used in our previous research (Goddard e t al., 1996). MEPS solutions were marked on two dimensions:

Relevant means: a quantitative measure describing the number of discrete steps which allowed the subject to move towards the goal (see Platt & Spivack, 1 9 7 5 ~ ) .

Effectiveness: a qualitative measure assessing the effectiveness of a solution on a seven-point scale (1 = not at all effective; 7 = very effective). The criteria used to measure effectiveness was adapted from the rating used by Marx e t al. (1992).

Scores were summed across the four problems to give each participant a total score for means and effectiveness. Two independent raters scored the MEPS for both dimensions, one of whom was blind to participants’ group membership. Pearson correlation coefficients were .87 for relevant means and .82 for effectiveness.

Each MEPS vignettes was read twice to the respondent by the experimenter. A lapse of 40 seconds was imposed for thinking about the problem and how to solve it. This standard time period for solution formulation was enforced in order to equate groups as added task pressure in the dual-task condition ma” otherwise have resulted in much longer problem-solving latencies. After the standard time period, parLicipants were cued by the experimenter to give their solution. Prior to the first MEPS problem, participants were instructed about the task requirements (i.e. to find the ideal strategy) and were further instructed to attend to the thoughts and images that they experienced while they were thinking about the problem and how to solve it. After giving each solution, participants reported the thoughts

618 L. Goddard, B. Dritscbel and A. Bwton

experienced during problem-solution formulation. These largely took the form of autobiographical memories and were rated according to the same criteria as used for the cueing task. Our previous research has found inter-rater agreement on the classification of these memories to be reliable.

The order of the two tasks (cueing task and MEPS) and the order of the four MEPS vignettes was alternated to avoid order effects. Practice trials were given on all tasks. In the case of dual-task participants, these practice trials were administered singularly.

Secondary tusks. These were computerized, four-choice reaction time tasks programmed using the Micro Experimental Laboratory (MEL; Schneider, 1990) package. In the easy dual task, one of four stimuli appeared in the centre of the screen for a duration of 1.5 seconds. The participants’ task was to depress a corresponding response key as quickly as possible. The stimuli were four characters on a qwerty keyboard and the response keys were specified keys of the keyboard. The response keys were labelled according to their allotted corresponding stimulus. The four stimuli were also present throughout the experiment at the bottom of the computer screen in the same spatial positions as those of the response keys.

If the participant did not respond within the maximum 1.5 seconds allowed, the target stimulus was replaced with another. The stimuli were presented in a random sequence. The task commenced with 20 practice trials, initiated by the participant on depression of the space bar. These were performed in the absence of the primary task. On completion of the practice trials, the secondary task was performed in conjunction with the primary task (i.e. the cueing task or MEPS). As the duration of these two primary tasks was determined by the participant, the number of trials performed on the secondary task differed between participants. For this reason the computerized task was designed with a large enough number of trials to ensure that its duration was compatible with the duration of the primary task for each participant. When all had completed the primary task, they were instructed to stop the computer task. In every case this was prior to the completion of the block of trials on the secondary task. In order to determine where in the secondary task data the participant had ceased to perform (i.e. had completed the primary task), the computer task data was examined for where the omissions from the end of the block began. These omission trials were then excluded from the reaction time and error analysis.

The difficult dual task was a modification of the above. In this task, the stimuli were no longer presented at a steady rate of 1.5 seconds, but were paced by the participant: each stimulus disappeared and was replaced on the participant’s response. In practice, the task therefore became faster as the maximum response time was 1.5 seconds. A further modification was a change in response keys. This was implemented by changing the spatial positions of the four stimuli at the bottom of the screen. This occurred at random intervals between 50 and 100 trials. Finally, in this task, participants received feedback on omissions and errors via a beep emitted from the computer. There were 40 practice trials with a response strategy change incorporated after the first 20 trials.

The secondary task was performed by the dual-task groups throughout the entire duration of the MEPS task and the cueing task. Participants were instructed to commence with the secondary task. After approximately 10 seconds of correct key responses, the first MEPS vignette was administered. Because participants may use different strategies in how they choose to divide their attention between the two tasks, they were instructed to direct most of their attention to the computer task. It was explained that this was the most important component of the experiment and that it was critical that they were as fast and as accurate as possible. In order to check that they had attended sufficiently and that there was minimal trade off between the tasks, error rates were checked by summing incorrect responses and omissions during performance. Four participants (two from each dual-task group) were replaced as their error rates were in excess of 10 per cent.

Results

Autobiographical Memory Cueing Task

Latency. Mean latencies (in seconds) to retrieve specific memories to positive and negative cues for females and males in all conditions are shown in Table 1. A 3 (task condition: single, easy dual, difficult dual) x 2 (gender: female, male) x 2 (valence:

Gender and social problem solving 619

positive, negative) mixed analysis of variance was applied to the data to examine times to retrieve a specific memory to cues. There were no main effects: task condition (F(2,54) = 2.53,p > .05, MSE = 122.85); gender (F(1,54) = 1.79,p > .l, MSE = 86.96); valence (F(1,54) = .78,p > .1, MSE = 15.52). However, there was a significant gender x valence interaction (F(1,54) = 5.14,p = .027, MSE = 102.80). Post hoc analyses (simple effects) found this to be due to the females (regardless of task condition) offering specific memories to negative cues significantly faster than males (F(1,54) = 4.97, p = .03). All other interactions were non-significant ( F < 1).

Table 1. Autobiographical Memory Cueing Task : Mean latencies (in seconds) to retrieve specific memories to cues (standard deviations are shown in parentheses)

Task condition Gender Positive cues Negative Total mean ~ ~

Single Female 11.4 (6.02) 9.74 (4.51) 10.57 (4.) Male 9.69 (5.07) 12.07 (4.7) 10.88 (3.54)

Female 9.91 (4.78) 10.74 (6.97) 10.33 (4.81) Male 8.21 (3.0) 13.61 (8.24) 10.91 (5.35)

Female 12.88 (7.54) 10.32 (6.22) 11.60 (6.67) Male 15.84 (5.69) 15.78 (5.58) 15.81 (4.59)

Easy dual

Difficult dual

SpeczFcity. The number of specific memories given as a first response to cues was analysed using a 3 x 2 x 2 (task condition x gender x valence) mixed analysis of variance. All main effects and interactions were non-significant ( F < 1). When participants failed to give a specific memory it was either because they were failing to recall any memory or because they gave an inappropriate general memory. These inappropriate general memories were analysed with a 3 (task condition) x 2

Table 2. Autobiographical Memory Cueing Task: Mean number of general memories (categoric/extended) offered as a first response to cues

Task condition Gender Categoric Extended Total general specific

Single Female .9 (1.61) .8 (.79) 1.7 (2.16) 8.3 (2.16) Male .6 (.84) .7 (.48) 1.3 (.82) 8.6 (.84)

Female 1.4 (1.84) .5 (.71) 1.9 (1.79) 8.0 (1.76) Male .5 (.71) 1.6 (1.17) 2.1 (1.1) 7.8 (1.03)

Female 1.4 (1.35) .4 (.7) 1.8 (1.62) 8.0 (1.6) Male 2.5 (3.0) .3 (.48) 2.8 (3.16) 7.1 (3.07)

Easy Dual

Difficult Dual

620 L. Goddard, B. Dritschel and A. Btlrton

(gender) x 2 (valence) x 2 (error type: categoric vs. extended) mixed analysis of variance. A main effect of error type (F(1,54) = 4.52, p = .038, MSE = 3.75) showed that errors were more likely to be categoric than extended. However, an interaction with task condition (F(2,54) = 5 . 4 8 , ~ = .007, MSE = 4.55) showed that the easy dual-task group made more extended errors than the other two groups (simple effects of contrasts: (F(2,54) = 4 . 2 8 , ~ = .019). However, this was modified by a further interaction with gender (task condition x error type x gender : F(2,54) = 3.88, p = .027, MSE = 3.22) which revealed this effect to be due to male participants alone (simple effects of contrasts: F(1,54) = 14.01, p < .OOOl). In turn, difficult dual-task males were more likely to make categoric errors than males of the other two groups (F(1,54) = 8 . 6 , ~ = .005). All other main effects and interactions were non-significant ( F < 1). See Table 2 for means and standard deviations for errors on cueing tasks.

Means-End Problem-solving Task

The mean number of relevant means generated and mean effectiveness scores are shown in Table 3. A 3 x 2 (task condition x gender) analysis of variance was used to examine the total number of means generated by participants on the MEPS. There was no main effect of task condition (F(2,54) = 2.79,p > .05, MSE = 63.35) and no main effect of gender (F(1,54) = .58, p > . l , MSE = 13.07). However, there was a significant task condition x gender interaction (F(2,54) = 3.28, p = .05, MSE = 74.32). Simple effects of contrasts found this to be due to the females in the single task condition who generated more relevant means than both the easy dual-task females (F(1,54) = 11.43, p = .OOl) and the difficult dual-task females (F(1,54) = 5.3, p = .025). Cognitive load did not, however, have any effect on males as performance did not differ between the groups. In order to examine gender differences, we also looked at male/female comparisons within groups ; here simple effects of contrasts showed single-task females to generate more relevant means than their male counterparts (F(1,54) = 6 . 2 , ~ = .016) whereas the remaining two groups did not show gender differences.

Table 3. Mean effectiveness score and mean total number of relevant means generated on the MEPS (standard deviations are shown in parentheses)

Task condition

Single Easy dual Difficult dual

Means Female 18.8 (7.0) 13.9 (5.17) 11.6 (2.76) Male 13.5 (2.76) 14.4 (4.43) 13.6 (5.04)

Female 19.9 (3.87) 20.8 (3.29) 15.7 (3.78) Male 19 (2.49) 20.4 (3.81) 15.3 (5.33)

Effectiveness

Gender and social problem solving 621

Analysis of variance was also used to examine MEPS effectiveness scores. This yielded a main effect of task condition (F(2,54) = 9.62, p < .0001, MSE = 143.12) which was due to the difficult dual-task condition where solutions were less effective than in the single-task and easy dual-task conditions (contrast: F(1,54) = 18.36,p < .0001. There was no main gender effect (F(1,54) = .32,p > .l, MSE = 4.82) and no task condition x gender interaction (F(2,54) = .03, p > .1, MSE = .42).

Memories retrieved during M E PS performance

The retrieval of specific memories during MEPS performance was analysed by summing the number of problems where participants retrieved a specific memory. Table 4 shows the mean number of problems where participants retrieved a specific memory during problem solving.

Table 4. Mean total number of specific memories retrieved during MEPS performance

Task condition

Single Easy dual Difficult dual

Female 2.3 (.82) .7 (.67) 1.9 (.94) Male .7 (.95) .9 (.57) .9 (.99)

The data were analysed using a task condition x gender analysis of variance. Both main effects were significant (task condition: F(2,54) = 3.85,p = .027, MSE = 2.72; gender: F(1,54) = 5.32, p = .025, MSE = 4.65). However, main effects were modified by a task condition x gender interaction (F(2,54) = 6.59, p = .003, MSE = 4.65). Simple effects of contrasts showed females in the single-task condition to retrieve more specific memories than females in both the easy dual-task condition (F(1,54) = 18.14, p = < .OOOl) and the difficult dual-task condition (F(1,54) = 11.98, p = .OOl). Again, within-group comparisons of gender showed females to retrieve more specific memories than males in the single-task condition (simple effects: F(1,54) = 18.14, p < .OOOl), whereas males/females in the dual-task conditions did not differ.

Having analysed the retrieval of specific memories during MEPS performance, we then analysed the occasions where participants had failed to retrieve a specific memory and had instead either failed to retrieve any memory or had retrieved categoric or extended general memories. These alternative retrieval strategies were analysed with a 3 (retrieval strategy: extended vs. categoric vs. no retrieval) x 3 (task condition) x 2 (gender) analysis of variance. As the main retrieval strategy effects of this analysis were identical to the reverse of examining specific retrieval during MEPS performance, the same main effects of task condition, gender and task condition x gender interaction were found and are explained, as above, in terms of

622 L. Goddard, B. Dritschel and A. Burton

females in the single condition reporting more specific memories and therefore fewer categoric/extended/no retrievals. The analysis of alternative retrieval strategies to specific retrieval found a main effect of alternative retrieval strategy (F(2,108) = 25.58, p < .0001, MSE = 24.77). Contrasts showed that when participants failed to retrieve a specific memory they were significantly more likely to report no retrieval than either categoric or extended (F(2,108) = 56.12, p < .OOOl).

In summary, the findings suggest that performance under the dual task significantly reduces the effectiveness of MEPS solutions. Retrieval of specific memories during the MEPS and the generation of means also decreases. In agreement with our predictions, however, these effects were specific to females.

Discussion

The aims of the study were to examine the effects of two levels of difficulty of a secondary task on (a) SPS ability as measured by MEPS performance, (b) memory retrieval during SPS and (c) the ability to retrieve specific memories to word cues. Level of secondary task difficulty was found to be an important factor in performance on the cueing task and MEPS. As predicted, secondary task pressure affected retrieval on the MEPS more than retrieval to word cues. This has methodological implications for autobiographical memory research. and suggests that the retrieval process may be influenced by the method through which it is examined. However, before discussing this in detail we shall first address the most prominent findings, i.e. those which concern gender differences in dual-task effects on SPS and memory retrieval.

Females performed more poorly when performing a task concurrently with a SPS task, in comparison to females performing the SPS task alone. Dual-task females also retrieved fewer specific memories during SPS than control females. In contrast, dual- task males did not differ from single-task males in these two dimensions. However, rather than this being due to dual-task males’ resilience, it was due to the superior performance of single-task females who outperformed their male counterparts. These findings are consistent with our predictions. There are, however, several possible explanations for the relatively poorer performance of males in the single task condition: (a) they may have been less motivated to generate detailed MEPS solutions and retrieve specific memories, (b) the MEPS vignettes reflect female rather than male concerns or (6) means-end generation does not reflect males’ SPS ability, and specific retrieval is more characteristic of females’ SPS style. In order to assess motivation we first examined reaction times and errors on the secondary task, on the grounds that there may have been differential trade-offs in performance. This did not seem to be the case; males’ and females’ performance did not significantly Furthermore (in contrast to our prediction), males generated solutions which were equally as effective as their female counterparts and therefore it seems unlikely that the MEPS reflect predominantly female concerns. The latter explanation is perhaps most appropriate, as it would seem that males and females, in the absence of working-memory pressure, possess distinct social cognitive problem-solving styles

* T tests were non-significant; reaction times: t = .54; errors: t = 1.5 (p > .1).

Gender and social problem solving 623

with females favouring a more detailed approach resulting in superior means-end generation. If this were the case it is perhaps not surprising that females would consequently favour the rich, detailed database which specific memories provide. We do not, however, suggest that males perform the MEPS without recourse to autobiographical memory even though in the absence of specific retrieval, reports of no retrieval were the most likely response. It seems likely that males may instead rely on more abstract rules in the form of general memories which are more difficult to introspect upon.

Level of task difficulty therefore proved to be an important factor when assessing the quality of MEPS solutions generated. The easy dual task interfered with females’ ability to generate relevant means; however, a more difficult task was required in order to interfere with the ability to generate effective solutions (for both males and females). We used two levels of task difficulty because we predicted that a relatively subtle reduction in resources would be necessary in order to reduce retrieval, during SPS, from the specific to general. As expected the difficult task tended to wipe out reports of retrieval altogether. Unfortunately, the easier task tended to do the same and did not achieve its aim in producing more general memories. Finding the optimum level of task difficulty is problematic. One needs to reduce resource capacity sufficiently for retrieval to change, while allowing adequate resources for introspection upon this change to be possible. A very precise task is necessary and further research exploring the effects of less demanding tasks on memory during SPS would be useful.

Although the secondary tasks had marked effects on MEPS performance (particularly for females), the cueing task findings are more complex. There are three criteria which have typically been used to assess performance on this task: (a) mean latency to retrieve specific memories to cues, (6) the number of specific memories retrieved as a first response to cues and (c) the types of errors offered in the absence of an initial specific response. Neither secondary task had an effect on the first two. For the easy dual task, this agrees with our prediction that retrieving specific memories to word cues would require a more demanding secondary task than the MEPS in order to interfere with normal retrieval patterns. We suggested that this was because the retrieval requirements of the cueing task are less demanding than that of the MEPS since the emotion word cues represent readily provided memory descriptions for the participant. However, we were surprised to find that the difficult secondary task also failed to reduce the number of specific memories retrieved or have any effect on response latencies, arguably a potentially sensitive measure of resource reduction. One explanation could be that the difficult task did not sufficiently reduce resources despite having been found to interfere with both means generation and effectiveness on the MEPS. Alternatively, participants may have compensated for their resource reduction by altering their retrieval strategy. One possible way in which this may have occurred is by processing the emotional concepts (i.e. the memory descriptions) less fully and/or shifting the criteria used to judge the suitability of memories retrieved in order to perform as quickly as possible. Thus the experiences retrieved by the dual-task groups may have been less appropriate exemplars of the cue words given than those of the controls. Another possible explanation for why the cueing task remained immune to the effects of the

624 L. Goddard, B. Dritschel and A. Burton

secondary task while the MEPS was more readily affected concerns the purpose of the end retrieval product. For the cueing task, retrieval requirements are satisfied simply by a match with the description. Apart from the task demands of specificity, verification of the target retrieved can be regulated by the participants and may be mediated by factors such as motivation and resource availability as described above. However, although the task demands of retrieval during the MEPS do not explicitly require specificity, they do require a target which provides the appropriate knowledge from which to define the given problem and generate an effective solution. This process may accommodate resource reduction by the retrieval of general information rather than specific.

It appears that the retrieval process involved in the cueing task may more readily accommodate secondary task pressure than the retrieval process involved during the MEPS. Nevertheless, it should not be assumed that latency and specificity would continue to remain resistant to secondary task demands if resources were more dramatically reduced. Indeed observation of the means, suggests that there is a trend towards longer latencies in the difficult dual-task condition (at least as far as males are concerned). Other response measures also suggest that the secondary tasks had an effect on performance. Although performing a secondary task did not increase the number of errors, there was a significant effect on the types of errors made by participants, for which difficulty proved to be a factor. Under the easy secondary task, males made more extended errors than other males, whereas under the difficult dual task, males made more categoric errors than other males. It is likely that extended memories may require more resources than categoric memories since extended retrieval on the cueing task has been found to correlate negatively with level of depression (Williams & Williams, 1996). The tendency for the easy/difficult dual-task conditions to give extended/categoric errors respectively may therefore reflect the relative resource demands of the two tasks. The error types of females were not affected by a secondary task. In view of the findings on the MEPS task (i.e. that with adequate resources, females are more likely to utilize specific memories than males), it is possible that females are less affected by a secondary task because the search strategies used to retrieve specific memories are more frequently rehearsed. Furthermore, females’ greater use of specific memories during SPS appears to be echoed in their faster latencies to retrieve specific memories than males’ (although this was to negative cues alone). This has implications for autobiographical memory research and suggests that individual differences warrant more attention.

It is interesting to contrast our findings with other research examining the effects of concurrent tasks on memory retrieval. Craik e t al. (1996) examined the effects of divided attention on the encoding and recall of word lists. In agreement with Baddeley e t al.’s (1984) earlier research, they found that divided attention at recall had little effect on performance whereas encoding was debilitated by resource reduction. In contrast our study shows that retrieval can be affected by a concurrent task. Clearly, the nature of autobiographical memory retrieval processes are very different from those involved in other non-autobiographical episodic memory tasks. Moreover, one important point which this study highlights is that retrieval from autobiographical memory itself may not be explicable in terms of a unitary process. The differential effects of the secondary tasks on the cueing and MEPS task show that

Gender and social problem solving 625

the method by which autobiographical memory is examined is an important factor in either the retrieval strategy employed or the retrieval demands it makes. Our results suggest that the retrieval strategies employed on the cueing task are resistant to reduced central executive capacity because neither of the secondary tasks reduced the ability to retrieve specific memories or increased latencies to retrieve specific memories (although difficulty level influenced the types of errors made). In contrast, retrieval during SPS appears to be less flexible in that it is highly dependent on sufficient capacity because both secondary tasks interfered with the ability to retrieve specific memories. Whether these differential retrieval effects imply that qualitatively different retrieval strategies are employed on the cueing task and MEPS or whether the effects are more accurately explained in terms of quantitatively different resource demands, is debatable and has not been addressed in this study. It is important to note, however, that the differential effects of the secondary tasks on the cueing and MEPS may well be specific to the visual four-choice reaction task employed in the current study. Indeed, secondary tasks which tap into different retrieval mechanisms may produce variable findings. For example, retrieval on the cueing task may be more readily affected by visual-spatial tasks as constructing memories to emotion word cues may have a more visual component than constructing memories for SPS purposes. In general the results endorse Bekerian & Dritschel’s (1992) view that ‘ Explanations used to account for memory under certain conditions should not be generalized to memory under different retrieval conditions ’. Our findings also suggest that methods other than the cueing task must be employed in order that a comprehensive theory may be developed.

In conclusion, this paper suggests that the central executive plays a major role in specific memory retrieval during the MEPS. However, the retrieval strategies involved on the cueing task appear more resistant to reduced central executive capacity (as measured by a four-choice reaction time task). It has been suggested that specific memories are important in SPS skill (Williams, 1996). Our findings suggest that this is true for females, but that males may use a more generic database for solving social problems. These results may have implications for research into gender differences in depression, as the data suggest that females experience a greater change in their SPS style (than males) when resources are reduced. Further examination of individual differences in autobiographical memory and SPS is clearly warranted. In particular, research with a personality/temperament scale could be a useful addition to the area as it may reveal another dimension accounting for gender differences.

References Baddeley, A. D. & Hitch, G. J. (1994). Developments in the concept of working memory.

Neuropsychology, 8, 485-493. Baddeley, A. D., Lewis, V., Eldridge, M. & Thompson, N. (1984). Attention and retrieval from long-

term memory. Journal of Experimental Psychology : General, 113 (4), 51 8-540. Baddeley, A. D. & Wilson, B. (1986). Amnesia, autobiographical memory, confabulation. In D. C.

Rubin (Ed.), Autobiographical Memory, pp. 225-252. Cambridge: Cambridge University Press. Bekerian, D. A. & Dritschel, B. H. (1992). Autobiographical remembering : An integrative approach.

In M. A. Conway, D. C. Rubin, H. Spinnler & W. A. Wagenaar (Eds), Theoretical Perspectives on Autobiographical Memoy, pp. 135-149. Dordrecht, Boston and London : Kluwer Academic.

626 L. Goddard, B. Dritschel and A. Barton

Conway, M. A. (1992). A structural model of autobiographical memory. In M. A. Conway, D. C. Rubin, H. Spinnler & W. A. Wagenaar (Eds), Theoretical Perspectives on Autobiographical Memory, pp. 167-193. Dordrecht, Boston and London: Kluwer Academic.

Craik, F. I. M., Govoni, R., Naveh-Benjamin, M. & Anderson, N. D. (1996). The effects of divided attention on encoding and retrieval process in human memory. Journal of Experimental Psychology: General, 125(2), 159-180.

D’Zurilla, T. J. & Goldfried, M. R. (1971). Problem solving and behaviour modification. Journal of Abnormal Psychology, 78, 107-126.

Engel, S. (1986). Learning to reminisce: A developmental study of how young children talk about the past. Unpublished PhD dissertation, City University of New York Graduate Center.

Evans, J., Williams, J. M. G., O’Loughlin, S. & Howells, K. (1992). Autobiographical memory and problem solving strategies of parasuicide patients. Psychological Medicine, 22, 399405.

Gilhooly, K. J., Logie, R. H., Wetherick, N. E. & Wynn, V. (1993). Working memory and strategies in syllogistic-reasoning tasks. Memory and Cognition, 21, 115-124.

Goddard, L., Dritschel, B. & Burton, A. (1996). Role of autobiographical memory in social problem- solving and depression. Journal of Abnormal Psychology, 105 (4), 609-616.

Goddard, L., Dritschel, B. & Burton, A. (1997). Social problem solving and autobiographical memory in depressed students. British Journal of Clinical Psychology, 36, 449451.

Grant, D. A. & Berg, E. A. (1948). A behavioural analysis of degree of reinforcement and ease of shifting to new responses in a Weigl-type card sorting problem. Journal of Experimental Psychology, 38, 404-41 1 .

Guttentag, R. (1989). Age differences in dual-task performance : Procedures, assumptions and results. Developmental Review, 9, 146-170.

Josephs, R. A., Markus, H. R. & Tafarodi, R. W. (1992). Gender and self-esteem. Journal of Personality and Social Psychology, 63, (3), 391402.

Kolodner, J. (1985). Memory for experience. In G. H. Bower (Ed.), Psychology of Learning and Motivation, vol. 19. Hillsdale, NJ: Erlbaum.

Logie, R. H., Gilhooly, K. J. & Wynn, V. (1994). Counting on working memory in arithmetic problem-solving. Memory and Cognition, 22 (4), 395-410.

Lykes, M. B. (1985) Gender and individualistic vs. collectivist bases for notions about the self. Journal of Personality and Social Psychology, 53, 356-383.

Marx, E. M., Williams, J, M. G. & Claridge, G. S. (1992). Depression and social problem solving. Journal of Abnormal Psychology, 101, 78-86.

Moscovitch, M. (1994). Cognitive resources and dual-task interference effects at retrieval in normal people: The role of the frontal lobes and medical temporal cortex. Neuropsychology, 8 (4), 524534.

Nelson, K. (1991). Remembering and telling: A developmental story. Journal of Narrative and Lije Histoy, 1 (2/3), 109-127.

Nolen-Hoeksema, S. & Girgus, J. S. (1994). The emergence of gender differences in depression during adolescence. Psychological Bulletin, 115, 424-443.

Nolen-Hoeksema, S. & Larson, J. (1992). The worries of adolescent males and females. Unpublished manuscript, Stanford University. Cited in Nolen-Hoeksema, S. & Girgus, J. S. (1993). The emergence of gender differences in depression during adolescence. Psychological Bulletin, 115,424443.

Norman, D. A. & Bobrow, D. G. (1979). Descriptions: An intermediate stage in memory retrieval. Cognitive Psychology, 11, 107-123.

Norman, D. A. & Shallice, T. (1980). Attention to action: Willed and automatic control of behavior. Technical Report No. 99. University of San Diego: Center for Human Information Processing.

Platt, J. J . & Spivack, G. (1975~). Manual for the Means-Ends Problem-Solving Test (MEPS): A Measure of Interpersonal Problem Solving Skill, Philadelphia : Hahnemann Medical College.

Platt, J. J. & Spivack, G. (1975b). Unidimensionality of the Means-Ends Problem Solving (MEPS) procedure. Journal of Clinical Psychology, 44, 15-16.

Reese, E. (1996). Conceptions of self in mother-child birth stories. Journal of Narrative and Lye History, 6 (l), 23-38.

Reese, E. & Fivush, R. (1993). Parental styles of talking about the past. Developmental Psychology, 29 (3), 596-606.

Gender and social problem solving 627

Reese, E., Haden, C. A. & Fivush, R. (1993). Mother-child conversations about the past: Relationships

Schank, R. C. (1982). Dynamic Memory. Cambridge: Cambridge University Press. Schneider, A. (1990). The Microexperimental Laboratoty. Pittsburgh : Psychology Software Tools. Sehulster, J. R. (1988). Broader perspectives in everyday memory. In M. Gruneberg, P. Morris & R.

Sykes (Eds), Practical Aspects of Memory: Current Research and Issues, vol. 1, pp. 323-328. Chichester, England: Wiley.

Sehulster, J. R. (1995). Memory styles and related abilities in presentation of self. American Journul of

Shallice, T. (1988). From Neuropsychology t o Mental Structure. Cambridge : Cambridge University Press. Sweller, J., Chandler, P., Tierney, P. & Cooper, M. (1990). Cognitive load as a factor in the structuring

Tannen, D. (1990). You Just Don’t Understand. New York: New York: Ballantine Books. Toms, M., Morris, N. & Ward, D. (1993). Working memory and conditional reasoning. Quarter&

Journal of Experimental Psychology, 46a (4), 679-699. Williams, D. M. & Hollan, J. D. (1981). The process of retrieval from very long-term memory. Cognitive

Science, 5, 87-1 19. Williams, W. H. & Williams, J. M. G. (1996). Autobiographical memory in severe head injury:

Influence of neurological damage and disturbance on recall. Proceedings of the British Pphology Sociep, vol. 3. no. 2.

Williams, J. M. G. (1992). The Psychological Treatment of Depression: A Guide to the Theory and Practice of Cognitive Behaviour Therapy. London and New York : Routledge.

Williams, J. M. G. (1996). Depression and the specificity of autobiographical memory. In D. Rubin (Ed.), Constructuring Our Past: A n Overview of Autobiographical Memory. Cambridge : Cambridge University Press.

Williams, J. M. G. & Dritschel, B. (1992). Categoric and extended autobiographical memories. In M. A. Conway, D. C. Rubin, H. Spinnler & W. A. Wagenaar (Eds), Theoretical Perspectives on Auto- biographical Memory, pp. 391412. Dordrecht, Boston and London : Kluwer Academic.

Woike, B. A. (1 992). A functional utility model of cognitive complexip. Unpublished doctoral dissertation, Michigan State University, East Lansing.

of style and memory over time. Cognitive Development, 8 (4), 403-430.

Psychology, 108 (l), 67-88.

of technical material. Journal of Experimental Psychology: General, 119 (2), 176192.

Received 26 June 1997; revised version received 26 February 1998