Research Repository and Digital Collections
Home of Minerva Access and Digitised Collections.
- Minerva Access is the University of Melbourne Institutional Repository and contains full text where available
- Digitised Collections contains digital copies of material from select Library Special Collections and other locations
Find items in two ways: search using the search bar or select a community below to browse.
Browse
Select a community to browse its collections.
Now showing 1 - 2 of 2
- Digital copies of resources from the University Library that are of cultural and historical significance to the University, including content from the Library's Special Collections.
- Minerva Access is the University of Melbourne’s institutional repository. Here you will find the open access research outputs of staff and students, including journal articles, books and chapters, and theses.
Recent Submissions
Perfect enough to sleep? Perfectionism and markers of poor sleep
Oh, Susie Yen Peng (2026)
Insomnia or poor sleep in the form of inadequate sleep either in duration or quality is a prevalent and costly health condition. Perfectionism, a multidimensional personality construct, has been theorised to be a risk factor of insomnia. Both insomnia and perfectionism are transdiagnostic risk factors in mental health disorders. While studies have investigated the relationship between perfectionism and an individual’s subjective perceptions of sleep (such as duration, quality and so forth), there is a lack of research looking at the relationship between objective measures of sleep and perfectionism; and none based on actigraphy sleep study. Furthermore, no previous research has investigated whether perfectionism plays a role in the commonly experienced insomnia symptom of sleep-wake state discrepancy, and whether sleep can also impact on perfectionism.
This thesis aimed to investigate the potential relationships between sleep and perfectionism and explore the mediating influence of stress, arousal and sleep-related cognitions on these relationships. In study 1, a comprehensive study of the relationship between sleep and perfectionism was conducted using subjective and objective sleep measures obtained from sleep diary and actigraphy, respectively. This study found self-driven, higher levels of internalised perfectionistic concerns about making mistakes and having doubts about actions were associated with poorer subjective sleep. In comparison, externally driven perfectionistic concerns of perceived standards that others have for oneself (socially prescribed perfectionism), parental expectations and parental criticism related positively to objective measures of poor sleep and stress mediated those relationships.
In Study 2, the potential role that perfectionism may play in sleep-wake state discrepancy was explored. It was found that higher levels of perfectionistic concerns over mistakes and doubts about actions correlated with and predicted discrepancy in sleep onset latency via the mediating effect of cognitive pre-sleep arousal. Study 3 then investigated the day-to-day bidirectional relationships between sleep and perfectionism at both interindividual and intraindividual levels. Bidirectional relationships between sleep and perfectionism were found at the interindividual level, but not at the interindividual level. However, stress mediated the relationship from perfectionism to sleep quality at the intraindividual level. Stress also mediated the relationship from sleep to perfectionism at the interindividual level.
Together, this thesis contributes new evidence to the existing literature that perfectionism is a risk factor of insomnia. It also highlights stress and pre-sleep arousal as underlying mediators in those relationships. These findings have important clinical and research implications in the assessment and treatment of poor sleep, suggesting that the different facets of perfectionism affect sleep differently and provide nuanced insights into how perfectionism affect sleep. In clinical practice, perfectionism should be assessed as a risk factor for insomnia. If perfectionism is identified as a potential driver of poor sleep, managing perfectionistic concerns, stress and pre-sleep arousal become important treatment considerations that can impact on the treatment process and therapy efficacy.
The Neural Mechanisms of Dietary Decision-Making
Chae, Jungwoo (2026)
Every day, people make numerous decisions about what to eat. Despite the frequency and personal significance of these choices, the neural and computational mechanisms that underlie dietary decision-making are not fully understood. This thesis aimed to characterise the mechanisms underlying two core stages of dietary decisions: the valuation phase, during which food attributes are appraised and represented in the brain, and the choice phase, during which these representations are converted into a decision outcome.
Chapter 1 provides the theoretical background for the thesis. I review traditional approaches in food cognition research and identify gaps that motivated the work in the empirical chapters. I outline key theoretical frameworks, including economic and mathematical theories of choice and neural models of decision-making, with a focus on the valuation and choice processes underlying dietary decisions. The chapter closes with the motivations for curating a large-scale, open-access neural and behavioural dataset.
Chapter 2 describes the experimental paradigms and analytical approaches used in the empirical chapters. I explain the rationale for the study design and stimuli curation, and introduce multivariate and event-related potential (ERP) analyses used throughout the thesis. Lastly, I describe the data preparation steps undertaken to create the open-access repository of neural and behavioural data.
Chapter 3 introduces the FOODEEG dataset, an open-access resource comprising electroencephalographic (EEG) responses from 117 participants to a diverse set of food images, accompanied by survey measures of food-related motivation and dietary styles and normative ratings on the food stimuli across 22 food attributes. The analyses in subsequent chapters of the thesis are performed on this dataset. This dataset descriptor constitutes a methodological contribution to the field, addressing the limited availability of publicly accessible and large-scale neural datasets in food cognition research. Released with standardised formatting, experimental code, and thorough documentation, the FOODEEG dataset is intended to support a wide range of future investigations into the neurocognitive processes underlying food-related behaviour, while adhering to open science principles of transparency and reproducibility. This chapter also provides a detailed account of the experimental design and dataset that form the basis of the empirical work in the subsequent chapters, and includes technical validation analyses of the EEG data.
In Chapter 4, we performed representational similarity analysis (RSA) on EEG data to characterise the neural time-courses of food attribute representations during valuation. Moving beyond the narrow focus on the trade-off between healthiness and tastiness that has dominated prior work, we examined a broad range of attributes spanning nutritive properties, hedonic properties, and familiarity. The majority of attributes were represented rapidly and in parallel, overlapping time windows. Contrary to longstanding assumptions in the field, healthiness was represented as early as, or earlier than, tastiness. Applying principal component analysis to address substantial covariation among attributes, we identified two broad underlying dimensions (how appetising and how processed a food is) represented in the neural signals with distinct time-courses, revealing a multilevel representational architecture. A subset of individual attributes contributed unique variance to the neural signals beyond this shared structure. Together, these findings reveal that food appraisal is a rapid, automatic process in which a diverse array of choice-relevant attributes is encoded simultaneously within the first second of viewing.
In Chapter 5, we investigated how these endogenous value representations are converted into an action outcome. Using a task designed to temporally separate attribute appraisal from choice and action execution, we examined whether the centroparietal positivity (CPP), an established neural marker of evidence accumulation in perceptual decision-making, exhibits similar dynamics during dietary decisions. We applied careful signal deconvolution to address methodological concerns regarding overlapping stimulus- and response-locked activity. We found that the CPP displayed hallmark signatures of evidence accumulation across three types of dietary decisions, including steeper pre-response slopes for trials with faster response times (RTs) and stronger representational strength, alongside a relatively fixed pre-response amplitude. Our findings support the idea that the CPP is a domain-general neural correlate of evidence accumulation, extending to dietary decisions. We reveal that evidence accumulation may underlie the choice process during different types of dietary decisions, including single-item willingness-to-eat decisions and dimension-specific attribute judgements of foods.
Chapter 6 provides a summary of the key findings across the empirical chapters and their implications. This thesis posits that a rapid, multilevel, and parallel food attribute appraisal process takes place during valuation and that noisy evidence accumulation underlies the conversion of these value representations into choice. These findings challenge traditional models of food choice that focus on the competition between automatic responses to palatability and effortful consideration of healthiness. I describe potential limitations across the studies and discuss how the contributions from this thesis can inform and advance research in clinical eating disorders and individual differences in dietary decision-making.
Together, these findings contribute to a mechanistically grounded, neurocognitive account of how people choose what to eat. A wide range of food attributes are appraised rapidly and at multiple levels during valuation, and these value representations are converted via a domain-general evidence accumulation process into choice outcomes. The open-access FOODEEG dataset supports future investigations into how these mechanisms are influenced by individual differences and contextual factors, and in clinical conditions. Taken together, this thesis brings us closer to building a process model of dietary decision-making.
Trust and risks in the Brazil-China soy economy
Zumpano Shelton, Petras (2026)
This thesis examines relations of trust and mistrust between stakeholders in the subsystem of the global soy economy that connects Brazil and China, including rural
producers, traders, regulators, activists and scientists. Building trust in the Brazil-China
soy economy is an enormous challenge due to the great diversity of identities and schools of thought. Transnational economic systems like the Brazil-China soy economy
are socially embedded and enmeshed in very diverse interpersonal and institutional networks of stakeholders. Risk perceptions, trusting dispositions and legitimacy beliefs
vary significantly. Conflicting risk perceptions and the inability of stakeholders to deal
with diversity generally undermine trust-building and legitimation processes in economic
life. This is both a conceptual and practical problem addressed by the thesis.
How is the Brazil-China soy economy socially embedded? That is the central question of the thesis. Asking this question can produce practical knowledge to build trust and legitimacy. Economic action is socially embedded in shared beliefs about trustworthiness and legitimacy that sustain habitual compliance with standards, norms
and agreements. Understanding the social construction of these beliefs in the Brazil
China soy economy can facilitate compliance and cooperation to improve governance.
The significance of this research question is tremendous given the scale of the Brazil-China soy economy, its importance for the global agricultural economy and its impact on
a great diversity of communities. More than one hundred research participants in Brazil
and China contributed to this project. Their input and the research results suggest that diversity must be acknowledged and understood for trust-building and legitimation to
occur.
Scholars and practitioners have published a vast body of literature on the conceptual and empirical associations of trust and legitimacy, but they often ignore or underestimate the importance of building interpersonal and institutional types of trust simultaneously. This thesis investigates how relations between states, markets and civil society in the Brazil-China soy economy are shaped by the intersection of trust and legitimacy. States and non-state institutions influence the social embeddedness of economic practices by guiding the formation of subjective meanings, personal identities and interpersonal trust between stakeholders, at the micro-social level. They also influence social embeddedness by building institutional trust and sharing, at the macro-social level, schemes of thinking about the legitimacy of governance. The soy economy provides evidence that trust and legitimacy intertwine across levels of analysis.
A Computational Analysis of the Structure and Evolution of Hierarchical Category Systems
Warburton, Katherine Ann (2026)
From folk taxonomies of animals to institutional systems like the Dewey Decimal Classification, hierarchical category systems shape how we think, communicate, and access information. Because these systems must compress vast amounts of data into meaningful but manageable categories, they inevitably reflect the needs, perspectives, and experiences of their creators. In this thesis, I present a large-scale computational analysis of the structure and evolution of hierarchical category systems, demonstrating how they encode bias through unequal representation of the things they classify, and how the order in which the environment is encountered can shape the structure and fit of category systems.
First, I examine gender bias in two institutional category systems, focusing on asymmetries in the existence of gendered subcategories and whether they reflect psychological principles of typicality. I compare these asymmetries to those in gendered language use, examining whether they can be explained by gender base rates and whether they have changed over time. Next, I develop a framework that conceptualizes how bias is encoded in the hierarchical arrangement of items and categories within a system. This framework measures Western and gender bias in two US-based library classification systems with millions of categories and is applied to animal concepts in English WordNet to demonstrate its broader applicability to cultural category systems. Finally, I propose that one reason hierarchical category systems can become biased is that they are formed incrementally with incomplete knowledge of the space of items they must categorize. Using a combination of computational modelling and behavioural experiments, I show that locally optimal decisions can produce suboptimal hierarchies depending on stimulus order. The model adapts a hierarchical clustering objective and outperforms existing rational models in predicting human behaviour, highlighting the need for intrinsically hierarchical models.
Overall, this thesis shows how hierarchical category systems can encode skewed representations of the world through both the biases embedded in their structure and the incremental processes by which they evolve. The methods developed here provide a foundation for future work on bias mitigation in both human and machine-generated classification systems and contribute to a better understanding of categorization in general.
Metabolic dysfunction-associated liver disease in Australia: epidemiology and therapeutic applications of medical weight management
Farrell, Ann (2026)
Metabolic dysfunction-associated steatotic liver disease (MASLD) is the most common cause of liver disease worldwide, with its rise mirroring the obesity epidemic. It represents a growing cause of morbidity and mortality, yet estimates of disease prevalence within Australia, and identification of those at risk of developing complications, remain lacking. Furthermore, therapeutic options are limited, with weight loss fundamental to management but difficult to achieve and maintain. This thesis explores the epidemiology of MASLD in Australia and investigates whether medical weight management approaches developed for obesity are effective, and whether they can be safely and effectively adapted for the treatment of MASLD.
To provide estimates of MASLD in Australia, a cross-sectional analysis of a large nationwide cohort (n=4,747) study was performed, and found elevated ALT in 13%, and MASLD in 37%. 61% of those with MASLD also had an elevated BARD score, suggesting that a large proportion of Australian adults with MASLD may warrant further investigation for risk of advanced fibrosis.
Real-world weight loss outcomes were examined in a retrospective cohort study (n=411) from a publicly-funded specialist weight management service. Using a very low energy diet (VLED) based approach, 50% achieved clinically significant weight loss (>= 10% total body weight loss (TBWL)) at 12 months, with a mean 10.5% TBWL. However, clinic retention was challenging, with 29% patient retention at 12 months. Metabolic phenotype (specifically presence of MASLD and insulin resistance) and early treatment response were found to be predictors of weight loss outcomes at 12 months.
The DARWIN randomised pilot study compared a 12-week ketogenic VLED to a Mediterranean diet (MD). The VLED achieved significantly greater hepatic steatosis reduction than a Mediterranean diet (77% vs 14% relative reduction), and resulted in greater weight loss (13% TBWL vs 4%). Low-dose semaglutide (0.5mg weekly) maintained these benefits to week 24 at which point superior histological improvement was also demonstrated in the VLED group. Extension follow-up to week 48 (24 weeks after cessation of all active intervention) found that the VLED group maintained clinically significant improvements in weight (10% TBWL) and hepatic steatosis (55% relative reduction), despite some weight regain and modest liver fat increase.
Collectively, these findings establish the substantial burden of MASLD in Australia, and demonstrated that the VLED-based medical weight management approach achieves clinically meaningful hepatic and metabolic benefits in MASLD. These results support the broader use of VLED-based programs as effective, accessible and a pragmatic first-line treatment for MASLD, but also highlight the need for improved engagement strategies in publicly-funded weight management services.