Meet Dr Raina Zhang: the researcher redefining how we understand memory and forgetting
What if forgetting was not the result of time erasing memories, but interference from new, similar experiences?
Dr Raina Zhang first encountered this “mind-blowing" theory of forgetting during her bachelors degree capstone subject with her supervisor, Associate Professor Adam Osth. Struck by how it challenged common assumptions, Raina’s interest in memory research was sparked.
“It motivated me to investigate the mechanisms behind memory success and failure – particularly why some items are easily remembered or misremembered than others,” she says.
Associate Professor Adam Osth’s expertise in memory research and computational modelling aligned perfectly with Raina’s research interests. The chance to continue working with him was thus a key reason why Raina decided to continue pursuing her research at the Melbourne School of Psychological Sciences.
“He has been an inspiring mentor since the capstone project who consistently supports both my academic development and scientific curiosity,” Raina explains.
Completing her entire higher education at the University of Melbourne – from undergraduate studies to her current postdoctoral position – Raina has benefitted from both consistent mentorship and a nurturing community within her hub.
I’ve come to deeply appreciate MSPS’s supportive environment and [the Complex Human Data Hub (CHDH)] in particular. Beyond excellent resources and working with world-leading experts in cognitive sciences, what makes CHDH special is how collaborative and welcoming it is. Between the weekly seminars, research development programs, casual hub morning teas and many other hub programs, it’s a place where everyone – from students to senior researchers – actively support each other’s growth.
Current research
Raina’s research asks fundamental questions about how we remember and misremember items, using computational models to bridge cognitive theory with real-world memory phenomena.
One of the primary areas of Raina’s research is refining word representation in memory models. When determining how people remember words, researchers often focus mainly on semantic representations (the meaning of words and how words relate to each other in definition and usage) using large language models. However, there is a lack of research on how perceptual and structural features (the word’s appearance and sound) can be represented in memory. These features include spelling (orthography) and pronunciation (phonology). During her PhD, Raina explored effective orthographic representations for memory models. Now she is extending this to phonology to build more comprehensive models of how word structure shapes memory.
Raina is also studying how robust false memories emerge – for example, when people confidently ‘remember’ words that were never actually presented, such as ‘sleep’ after exposure to related terms like ‘bed’, ‘rest’ and ‘awake’. She aims to build a computational model that explains when and why these false memories occur.
The third key area of her research is examining learning mechanisms in memory, investigating whether memory strengthens gradually with increased exposure or if learning reflects an ‘all-or-nothing’ process.
To test these theories, Raina conducts empirical experiments with specific manipulations. She then applies hierarchical Bayesian models to individual participants’ data. This approach not only reveals where models succeed or fail, but also enables comparison between different models to identify which model provides the best account of data.
The joy of discovery
What Raina enjoys most about her research is the process of applying mathematical and computational models to understand memory.
“There’s something deeply satisfying about taking abstract theories and translating them into testable predictions – then seeing how these predictions hold up against real data,” she reflects.
“Of course, it’s not always straightforward,” Raina admits. “It can be really frustrating when model predictions are off. But that’s also where the real learning happens. Troubleshooting forces me to dig deeper into the theory, rethink assumptions, and often leads to unexpected insights. And when everything finally clicks – when the model not only fits the data but reveals something new about how memory works – that's the most rewarding part.”
Ultimately, the interplay between theory, modelling and data is what keeps Raina engaged.
“It’s like solving a puzzle where the pieces are constantly reshaping themselves, and that intellectual challenge is what I love most about my work.”