Spatial Working Memory
Spatial Working Memory was a developed task that aimed to measure one’s ability to mentally rotate and recall dynamic spatial information. This task was
presented in the form of a maze with a cat in the centre. Participants viewed a dynamic line drawn from the centre of the maze out. Once the line disappeared the maze rotated 90, 180 or 270 degrees and participants were required to recall the correct path in the new orientation. The revision for Spatial Working Memory was born from inconsistent findings in the item analysis. All the developed span and working memory tasks were designed to follow the format of the reliable and valid span assessment paradigm (Strauss et al., 2006; Wechsler, 2008). The span assessment paradigm involves the presentation of two trials at each level of difficulty (span). The Spatial Working Memory task was initially designed to exactly mirror this presentation, with two presentations of rotated mazes at each span, however, three types of rotations were possible (90, 180 and 270 degrees). These three orientations were randomised to be presented in pairs across trials. The randomisation aimed to ensure that the participant would be unable to identify a pattern of rotation, and thus make accurate predictions about what rotation would be presented next, potentially inflating their Spatial Working Memory score. In an attempt to remediate this error and to ensure that every trial was of equal difficulty, each span on the revised Spatial Working Memory assessment was expanded to three trials, with each span containing one trial for each rotation (90, 180 and 270 degrees). Within each trial the three rotations would be presented in a
predetermined, randomised order. This means that all participants would receive the same trials presented to them in the same order however, this order was not be based on any pattern and therefore, the participant is unable to predict what stimuli will be
presented next which may have had the ability to enhance their spatial working memory score.
Object Working Memory
Object Working Memory was a developed task that aimed to measure one’s ability to manipulate and recall static, object information. The item analysis from the Object Working Memory task yielded expected findings, with scores within spans producing similar results and as span increased a decline in performance was observed. The issue that arose in this task regarded the construct validity being questioned, as findings did not conform sufficiently to the established theoretical notions of working memory. Unexpectedly, no differences were found between the forwards (span) and backwards (working memory) presentations of the stimuli. The decision to incorporate the same stimuli in both the span and working memory tasks was a notion derived from the Digit Span subtest of the WAIS-IV (Wechsler, 2008). Digit Span, whose design and administration follows the same theoretical underpinnings as the developed tasks, uses the numbers from 1 - 9 in both the forwards and backwards trials. The manual
highlights that numbers were chosen as the stimuli for this task as they carry no context or semantic meaning to individuals. As reiterated numerous times, it is known that verbal and visual information are inherently different, and this finding highlights that mirroring the stimuli design of a verbal memory task may not be appropriate when attempting to measure visual memory. Similar to Digit Span, the stimuli chosen for presentation in this task included abstract components designed to limit contextual or
semantic meaning. Despite attempts to limit context or meaning for the stimuli
presented, seeing the same stimuli presented in the span version of this task, may have allowed the individual to encode the different objects effectively (by creating verbal analogues) and to generate stored representations in their long term memory. Thus, by the time Object Working Memory was administered using the same stimuli, rather than having to hold the information, encode it and manipulate it, the individuals were likely able to recognise the object from their stored representations. As a result, the working memory component of this task possibly required reduced cognitive demand when compared to the span component. This notion of object familiarity aiding in recall has been established in the literature. Xie and Zhang (2017) identified that participants were readily able to recall objects that were familiar to them much more swiftly than
unfamiliar objects. This lead to participants demonstrating a greater capacity for objects that were known to them compared to unknown.
Based on this review, Object Working Memory was redesigned using different stimuli than Object Span. Therefore, each task should now involve the same perceptual processes where a participant is required to observe, develop an encoding strategy and hold the individual objects as they are presented, thereby attempting to standardise cognitive load across both tasks, asides from the additional manipulation component in the working memory component. Despite threats to construct validity arising from the design, Object Span had demonstrated fair internal consistency. In order to maintain this, each stimuli developed for use in the revised Object Working Memory task was matched to a stimulus in the original Object Span task, in terms of the number of features to be encoded (see Appendix M). Some studies contend that once an object is bound the number of features becomes irrelevant (Luck & Vogel, 1997), while more recently it has been found that an increase in the number of features makes an object
more difficult to bind (Taylor et al., 2012). To control for these inconclusive findings an attempt to standardise the number of features present in each object was made.
Similarly to Object Span, an effort was made to allow each stimulus designed to be easily identifiable as a stand alone object that was not able to be easily confused with any other objects that are presented.
Investigation of Performance Between the Spatial and Object Memory The primary purpose of Experiment 2 was to investigate if differences in performance exist between spatial and object memory. The notion that differences would be present within these two visual memory functions is based off the two streams hypothesis understanding of visual perception (Milner & Goodale, 1992). Past research that has investigated visual memory has consistently produced varied findings (Brady et al., 2011; Corsi, 1972; Pickering, 2001). Researchers have reported that there are
profound difficulties with replicating findings in scenarios where the stimuli are slightly altered (Humphreys, 2016). Furthermore, there is longstanding significant research that indicates how an individual processes and stores information pertaining to a visual scene (Green & Oliva, 2009) varies substantially to how an individual encodes specific objects (Brady et al., 2008). Moreover, many studies that have produced these findings have done so by incorporating both spatial and object stimuli (e.g. tasks exploring how object locations are stored) (Alvarez & Oliva, 2008). While understanding how the spatial and object information are processed together has high ecological validity, as in reality they often work in conjunction with one another, to understand the visual
memory system as whole, investigating how memory for spatial and object information is stored independently is also vital for discovering the influence that each stream has on visual memory function. Past research has investigated the functions independently
(Humphreys, 2016; Luck & Vogel, 1995; Ishikawa & Montello, 2006; Poirel et al., 2011), however, research that incorporates methods that isolate each domain, while designed in a manner that allows for direct comparison is scarce. Subsequently, after reviewing the psychometric properties that aimed to isolate spatial and object memory functioning, analyses were undertaken to determine whether a difference in capacity was present between the two streams.
Aims and Hypotheses
Experiment 2 aimed to investigate whether there were capacity differences between spatial and object memory. The aims, results and discussion for this experiment will be presented in two distinct sections - piloting the revised working memory
assessment tasks followed by an investigation of the capacity differences between the developed spatial and object memory tasks.
Experiment 2 involved piloting the revised working memory tasks. With that notion the psychometric properties (construct validity, internal consistency and internal validity) of these tasks was measured using the same construct parameters as outlined in Chapter 6.
Furthermore, based on the promising results of Experiment 1, a more exhaustive analysis of convergent and discriminant validity was conducted utilising specific
exemplar tests that model the memory assessment paradigms incorporated into the developed tests. Specifically;
• Performance on the developed span and working memory tasks would yield weak - moderate correlations with Digit Span.
• Performance on the developed learning tasks would yield weak - moderate correlations with the RAVLT.
Experiment 2 also involved investigating whether there were capacity differences between spatial and object memory.
It was therefore hypothesised that:
• Performance differences will be present between spatial and object tasks.
Method Participants
Experiment 2 involved the recruitment of a new sample of participants. As the present study incorporated revised versions of the working memory tasks as well as all other developed tasks, to reduce the impact of practice effects and task familiarity and to increase the likelihood of meaningful, unbiased results a new sample was required. To maintain integrity and consistency between the findings established by the former sample and the newly recruited sample, participants for Experiment 2 were recruited from the same regions of metropolitan Melbourne, Victoria, Australia using the same methods as those recruited in Experiment 1. Data was collected from 63 individuals, aged between 18 -45 (mean = 29.19, SD = 8.09). Table 7.1 displays a summary of the demographic data about the sample recruited for Experiment 2.
Table 7.1. Percentage of Participants Represented in Each Category for Nominal Demographic Data (N=63)
The sample sourced for Experiment 2 were of a similar demographic to those who participated in Experiment 1. The proportion of participants from each biological sex category was nearly the same. In terms of employment, more participants from this sample were full time workers which is contrary to the sample from Experiment 1 that saw a higher proportion of participants in the casual/part time category (likely due to the higher proportion of university students present in Experiment 1). In terms of level of education the present sample indicated that they were slightly less educated than the initial. While they has slightly less high school graduates and more individuals with a TAFE diploma, they had a smaller proportion of undergraduate degree holders and no one with a post graduate degree.
Table 7.2 shows the means, standard deviation and ranges for the sample’s performance on measures of IQ and established measures of memory.
Category Percentage of Participants in Each Group
%
Sex Female Male
55.60% N = 35
44.40% N = 28
Employment Unemployed Casual / Part
Time Full Time 20.60%
N = 13 34.90%N = 22 44.40%N = 28 Highest Level
of Education Year 11 Completion Tafe Certificate High School Completion Tafe Diploma
Undergrad Degree
7.90%
Table 7.2. Descriptive Statistics for Index Scores on the WMS and WASI (N=63)
Skewness and kurtosis indicate that all variables fell within the accepted range (-3 to 3) which indicates that the sample is normally distributed (George & Mallery, 2010). The Wechsler administration manuals outlines that the mean for all indices in the WMS and WASI is 100 with a standard deviation of 15 (Wechsler, 2009; Psychological Corporation, 1999). It can be seen from the above table that the sample means and SDs fall within range for all indices, excluding the standard deviations for the Visual
Working Memory Index and the Delayed Memory Index. Standard deviations for these variables fell outside the expected range, indicating high rates of variability, however, mean scores were as expected. Somewhat concerning are the low scores observed in the Visual Memory Index and the Visual Working Memory Index, in particular mean scores for the Visual Working Memory Index are lower than would be expected in comparison to other scores. Nonetheless these scores still fall squarely within one standard deviation of the mean. In comparison to the initial sample, in general this sample scored half a deviation lower on measures of intelligence and memory. This indicates on most subtests the present sample yielded slightly lower scores. While this slight difference is
Index Mean (SD) Range
Full Scale IQ 104.87 (12.68) 79 - 140
Verbal Comprehension Index 104.52 (13.43) 79 -140 Perceptual Reasoning Index 105.80 (13.24) 77 - 131 Auditory Memory Index 103.75 (14.05) 80 -136 Visual Memory Index 97.84 (15.76) 57 - 134 Visual Working Memory Index 87.72 (10.87) 60 - 120 Immediate Memory Index 102.88 (13.97) 80 - 132 Delayed Memory Index 103.14 (17.98) 67 - 141
present, this data indicates that the sample is normally distributed and representative of the pattern of performance seen in the general population.
Materials
As highlighted in the test revision section, Spatial Working Memory and Object Working Memory were revised prior to the commencement Experiment 2. Based on the findings of Experiment 1 Spatial Working Memory has been adapted and now includes an extra trial for each span (making three trials total) to ensure each span is equivalently difficult in terms of stimuli rotation. Also based on the findings of Experiment 1 Object Working Memory now incorporates different stimuli than Object Span (as seen in Appendix M.). All other materials were unchanged from Experiment 1 and thus, details of each measure can be found in Chapter 6.
Experiment 2 while presented in sequence, utilises the same sample as the future Experiment 3. During Experiment 3, some methodological changes were implemented to provide a more exhaustive analysis of construct validity. Namely, the inclusion of two assessment tasks that align with the same administration paradigms utilised in the development of the spatial and object tasks. While a verbal memory task, Digit Span incorporates the well established span paradigm, that involves the inclusion of two trials to measure each span, and similar start and discontinue rules as the developed span and working memory tasks. Similarly the RAVLT is considered to be an exemplar
assessment that incorporates the list learning paradigm. Again, while verbal in nature it incorporates the same administration procedures as the developed learning tasks. While verbal memory is inherently different from visual memory they are both components of the same construct. While differences in capacity between these tasks are expected, relationships are also expected. Similarly, to how IQ is associated with memory, it