In this thesis, a discrete Markov chain model for analyzing memory interference in multiprocessors, is presented.

The present thesis aimed to investigate the source of interference in recognition memory. There are two primary alternatives – the item noise approach, in which interference comes about as a consequence of the other items on the study list, and the context noise approach, wherein interference arises from the previous contexts in which an item has been encountered. Alternatively, interference may occur through a combination of both item and context noise. There are many mathematical models designed to simulate the recognition process that incorporate either item or context noise, or both. Item noise models predict a significant list length effect, that is, that memory for an item that was part of a short list at study is better than that for an item that was part of a long list. Context noise models no not predict a significant difference in memory based on the length of the study list. The presence or absence of the list length effect can therefore be used as a mechanism by which to differentiate item and context noise models. The list length effect is among the most documented and replicated findings in the recognition memory literature (Gronlund & Elam, 1994). Yet, while many experiments have been conducted which have identified a significant list length effect in recognition (e.g. Bowles & Glanzer, 1983; Cary & Reder, 2003; Murnane & Shiffrin, 1991; Ohrt & Gronlund, 1999; Strong, 1912; Underwood, 1978), a number of published studies have failed to identify the effect (e.g. Dennis & Humphreys, 2001; Dennis, Lee & Kinnell, 2008; Jang & Huber, 2008; Murnane & Shiffrin, 1991; Schulman, 1974). Dennis and Humphreys (2001) argued that studies that had identified a significant effect of list length on recognition performance had done so because of a failure to control for four potentially confounding variables; retention interval, attention, displaced rehearsal and contextual reinstatement. The possible confounding effects of retention interval and displaced rehearsal are already well established in the literature and most studies employ some controls for these. Dennis et al. (2008) also found that while the role of contextual reinstatement had a pronounced influence on the detection of the list length effect it did not appear to be the most influential of the potential confounds. Thus, a major aim of the present thesis was to investigate the role of attention in the identification of the list length effect. Experiment 1 (N=160) involved two manipulations of attention. The first was to use either a retroactive or proactive design, with differential lapses of attention likely to be more pronounced in the latter. Second, in one condition participants were asked to perform a pleasantness rating task at study, a common technique to encourage participants to attend to the stimulus, while in the other condition they were asked to simply read the words. Results indicated that attention modulates the list length effect and that it is the retroactive versus proactive distinction which is most important as a significant effect of list length was found only when the proactive design was used. The encoding task had little effect. The design of Experiment 2 (N=80) was based on Cary and Reder's (2003) Experiment 3 which itself was a partial replication of Dennis and Humphreys' (2001) experiments. Cary and Reder introduced the Remember-Know (RK) task into the test list in their experiments and identified a significant effect of list length in the presence of controls for the four confounds where Dennis and Humphreys had not. The RK task is thought to index the relative contributions of familiarity and recollection in the recognition process (Gardiner, 1988). To the extent that the RK task encourages a recall-like process (see Clark, 1999; Diana, Reder, Arndt & Park, 2006) it may influence the results regarding the list length effect, in that the effect is widely accepted to occur in recall. Experiment 2 compared recognition memory with or without RK instructions. One condition involved the standard yes/no recognition paradigm, while the other made us of the RK task following all “yes” responses. Controls for the four potential confounds of Dennis and Humphreys were implemented. No significant effect of list length was identified in the accuracy data of either condition, however there was a small but significant effect on median response latency for correct responses in the RK task condition. The results of Experiments 1 and 2 suggest that the effect of list length on recognition performance is negligible and nonsignificant when controls for the four potential confounds of list length are in place. However, both of these experiments, and almost all previous experiments investigating the list length effect, used words as the stimuli. The remaining four experiments in the present thesis (N=40 in each) sought to investigate the list length effect using stimuli other than words in an attempt to identify the boundary conditions of the effect. Each of these experiments followed the same basic method as Experiments 1 and 2. Four different kinds of stimuli were investigated, word pairs, images of novel faces, fractals and photographs. Results indicated a nonsignificant effect of list length for word pairs and photographs, however, there was a significant list length effect when faces (in the accuracy data) and fractals (in the response latency data) were used as the stimuli. However, all of the experiments in the present thesis used a within subjects manipulation of list length in order to maximise experimental power. This design may be an additional confound of the list length effect. The nature of the within subjects design means that by the end of the second study list, all participants will have studied the same number of items, thereby potentially removing any list length manipulation from the experiment. In addition, participants who studied the long list first may be more likely to be affected by lapses in attention than participants who began with the short list with this, rather than interference, the potential cause of any list length differences. In order to investigate this potential confounding, the results from all experiments of the present thesis were re-analysed using a between subjects analysis based on only the first list studied by each participant. The qualitative conclusions drawn from the majority of conditions remained unchanged. The between subjects analysis generally revealed larger effect sizes than did the within subjects analysis, although with the exception of the proactive conditions, these effects can be considered negligible to small at most. The pattern of results across the six experiments of the present thesis are problematic for existing mathematical models of recognition memory. While context noise models are able to account for negligible and nonsignificant effects of list length when words, word pairs and photographs are used as the stimuli, they are unable to predict a slightly larger and significant list length effect when the stimuli are novel faces or fractals. Conversely, while item noise models are able to account for a significant list length effect for faces and fractals, they are unable to predict a nonsignificant list length effect for words and word pairs. The results question whether either item or context noise can be taken as the sole source of interference in recognition memory. Rather, a combination of interference from different sources may be at work, with the precise nature of this combination dependent on the nature of the stimuli involved. However, it is important to note that these models must be able to all but eliminate interference from other items under certain conditions to obtain the negligible list length effect findings reported here. / http://proxy.library.adelaide.edu.au/login?url= http://library.adelaide.edu.au/cgi-bin/Pwebrecon.cgi?BBID=1474563 / Thesis (Ph.D.) -- University of Adelaide, School of Psychology, 2009

abstract: The availability of a wide range of general purpose as well as accelerator cores on modern smartphones means that a significant number of applications can be executed on a smartphone simultaneously, resulting in an ever increasing demand on the memory subsystem. While the increased computation capability is intended for improving user experience, memory requests from each concurrent application exhibit unique memory access patterns as well as specific timing constraints. If not considered, this could lead to significant memory contention and result in lowered user experience. This work first analyzes the impact of memory degradation caused by the interference at the memory system for a broad range of commonly-used smartphone applications. The real system characterization results show that smartphone applications, such as web browsing and media playback, suffer significant performance degradation. This is caused by shared resource contention at the application processor’s last-level cache, the communication fabric, and the main memory. Based on the detailed characterization results, rest of this thesis focuses on the design of an effective memory interference mitigation technique. Since web browsing, being one of the most commonly-used smartphone applications and represents many html-based smartphone applications, my thesis focuses on meeting the performance requirement of a web browser on a smartphone in the presence of background processes and co-scheduled applications. My thesis proposes a light-weight user space frequency governor to mitigate the degradation caused by interfering applications, by predicting the performance and power consumption of web browsing. The governor selects an optimal energy-efficient frequency setting periodically by using the statically-trained performance and power models with dynamically-varying architecture and system conditions, such as the memory access intensity of background processes and/or coscheduled applications, and temperature of cores. The governor has been extensively evaluated on a Nexus 5 smartphone over a diverse range of mobile workloads. By operating at the most energy-efficient frequency setting in the presence of interference, energy efficiency is improved by as much as 35% and with an average of 18% compared to the existing interactive governor, while maintaining the satisfactory performance of web page loading under 3 seconds. / Dissertation/Thesis / Masters Thesis Electrical Engineering 2016

Several lines of evidence from rapid serial visual presentation, attentional blink, and dual-task interference phenomena propose that human beings have a significant limitation on the short-term consolidation process. Short-term consolidation is transferring early representations to more durable forms of memory. Although previous research has shown that masks presented after targets interrupt the consolidation process of information, there is not enough evidence for the role of attention in consolidation for episodic memory. One electrophysiological and five behavioral experiments were conducted to investigate the effects of attention and stimulus onset asynchrony (SOA) between targets and masks on episodic memory. Masks were presented after targets with varying SOAs. The participants in the divided attention condition but not the ones in the full attention condition performed the attention-demanding secondary task after the presentation of the masks. The results showed that reducing SOA between targets and masks caused an impairment in memory performance for divided attention but not for full attention, providing evidence for the necessity of attention for the short-term consolidation process. Electrophysiological results demonstrated that this impairment did not result from perceptual processes.

This research proposes that a construct called Sponsorship Linked Advertising (SLA) is valuable in understanding how brand and corporate advertising link to sponsorship and event marketing. SLA includes both ads that communicate a link to a sponsored event (tiedness) and those that demonstrate, through their overall design, an event’s motif or theme (themedness). In its themed form, SLA differs from creative advertising by virtue of the fact that creative advertising can exist independent of any sponsorship link, whereas sponsorship linked advertisements intentionally unite a sponsorship and an event either implicitly and/or explicitly. With sponsorship investment estimated to be $45.2 billion worldwide (International Events Group 2009) and leverage advertising (i.e., advertising that is employed to heighten awareness of sponsorships or better articulate sponsor-event links) reflecting a similar amount, empirical examination, validation and implementation guidance of SLA as a leveraging strategy is critical. The unique marketing opportunities associated with popular sporting, charitable and arts events also attract non-sponsoring companies which also seek to affiliate with the event, an activity known as ambushing (McKelvey and Grady 2008). Examination of SLA is therefore important in an increasingly competitive and cluttered global sponsorship arena, in which ambushing is becoming a pervasive practice, attracting considerable attention from event organisers, sponsors and policy makers alike, while also detracting from or diluting intended sponsorship communications. Despite widespread use of SLA, no empirical validation of this strategy has been undertaken to date. The present research addresses this gap. Initially, the SLA construct is defined, categorised, and measured through content analysis, then a series of experimental studies are used to achieve empirical validation of the SLA construct. Streams of sponsorship and advertising effectiveness research, along with theories of associative learning, attribution and persuasion, are used to guide examination of SLA effectiveness in new empirical work. Specifically, a series of experiments are used to examine consumers’ cognitive, affective and behavioural responses to SLA. The first experimental study tests a main effect of ad type as well as interactive effects of tiedness and themedness with sponsorship availability (i.e., knowledge of true sponsor) on outcomes including nature and type of thoughts elicited, ad scepticism and sponsor motive attributions. Findings from this study suggest that SLA exposure induces more positive thoughts, less ad scepticism and more favourable attributions than exposure to Non SLA. These results provide initial evidence that consumers process SLA differently to other ad types and interestingly, may derive enjoyment from, or at least exhibit less ad scepticism towards viewing SLA. Finally, the influence of competitive context on sponsor recall is tested by simulating exposure to ambush and/or SLA ad types following a sponsorship announcement. Findings provide evidence to support a memory interference hypothesis and imply that accuracy of sponsor recall is diminished by presence of an ambush ad, but that this effect is moderated by presence of SLA leveraging a previously announced sponsorship. Taken together, the results of this series of studies provide an empirical measurement of SLA strategy and demonstrate construct validity. Further, interpretation of the results gives rise to specific creative strategies for practical implementation. Ecological validity is built into the design by using real events within ad stimuli and investigating SLA in the competitive context in which it occurs. Hence, the results are said to be generalisable to real-world situations, and the resultant creative strategies are arguably contextually valid. This research contributes to existing marketing and sponsorship literature by proposing and empirically validating a new construct. Theoretically, it examines consumer response to SLA by combining information processing and resistance based perspectives. It extends traditional views of ambushing by offering empirical evidence of the practice being widespread and extending to low level sponsors and event “free riders”. Practical implications of this research extend to advertisers and sponsors faced with the challenge of effectively leveraging huge sponsorship investments and assessing return on such investment. Empirical testing of ambushing effects has important implications for the debate on increased regulatory intervention of such practices, a debate centred upon tension between balancing fair marketing practice with the rights of sponsors and event organisers.

This research proposes that a construct called Sponsorship Linked Advertising (SLA) is valuable in understanding how brand and corporate advertising link to sponsorship and event marketing. SLA includes both ads that communicate a link to a sponsored event (tiedness) and those that demonstrate, through their overall design, an event’s motif or theme (themedness). In its themed form, SLA differs from creative advertising by virtue of the fact that creative advertising can exist independent of any sponsorship link, whereas sponsorship linked advertisements intentionally unite a sponsorship and an event either implicitly and/or explicitly. With sponsorship investment estimated to be $45.2 billion worldwide (International Events Group 2009) and leverage advertising (i.e., advertising that is employed to heighten awareness of sponsorships or better articulate sponsor-event links) reflecting a similar amount, empirical examination, validation and implementation guidance of SLA as a leveraging strategy is critical. The unique marketing opportunities associated with popular sporting, charitable and arts events also attract non-sponsoring companies which also seek to affiliate with the event, an activity known as ambushing (McKelvey and Grady 2008). Examination of SLA is therefore important in an increasingly competitive and cluttered global sponsorship arena, in which ambushing is becoming a pervasive practice, attracting considerable attention from event organisers, sponsors and policy makers alike, while also detracting from or diluting intended sponsorship communications. Despite widespread use of SLA, no empirical validation of this strategy has been undertaken to date. The present research addresses this gap. Initially, the SLA construct is defined, categorised, and measured through content analysis, then a series of experimental studies are used to achieve empirical validation of the SLA construct. Streams of sponsorship and advertising effectiveness research, along with theories of associative learning, attribution and persuasion, are used to guide examination of SLA effectiveness in new empirical work. Specifically, a series of experiments are used to examine consumers’ cognitive, affective and behavioural responses to SLA. The first experimental study tests a main effect of ad type as well as interactive effects of tiedness and themedness with sponsorship availability (i.e., knowledge of true sponsor) on outcomes including nature and type of thoughts elicited, ad scepticism and sponsor motive attributions. Findings from this study suggest that SLA exposure induces more positive thoughts, less ad scepticism and more favourable attributions than exposure to Non SLA. These results provide initial evidence that consumers process SLA differently to other ad types and interestingly, may derive enjoyment from, or at least exhibit less ad scepticism towards viewing SLA. Finally, the influence of competitive context on sponsor recall is tested by simulating exposure to ambush and/or SLA ad types following a sponsorship announcement. Findings provide evidence to support a memory interference hypothesis and imply that accuracy of sponsor recall is diminished by presence of an ambush ad, but that this effect is moderated by presence of SLA leveraging a previously announced sponsorship. Taken together, the results of this series of studies provide an empirical measurement of SLA strategy and demonstrate construct validity. Further, interpretation of the results gives rise to specific creative strategies for practical implementation. Ecological validity is built into the design by using real events within ad stimuli and investigating SLA in the competitive context in which it occurs. Hence, the results are said to be generalisable to real-world situations, and the resultant creative strategies are arguably contextually valid. This research contributes to existing marketing and sponsorship literature by proposing and empirically validating a new construct. Theoretically, it examines consumer response to SLA by combining information processing and resistance based perspectives. It extends traditional views of ambushing by offering empirical evidence of the practice being widespread and extending to low level sponsors and event “free riders”. Practical implications of this research extend to advertisers and sponsors faced with the challenge of effectively leveraging huge sponsorship investments and assessing return on such investment. Empirical testing of ambushing effects has important implications for the debate on increased regulatory intervention of such practices, a debate centred upon tension between balancing fair marketing practice with the rights of sponsors and event organisers.