Acquiring expertise in the mental manipulation of visual images: Effects on brain and behaviour.

University of Newcastle Strategic Pilot Research Grant
Chief Investigators: A.Heathcote, S.D. Brown & K. Sutton
Associate Investigator: B. Johnson

The ability to rotate an imagined object ("mental rotation") is fundamental to many aspects of human functioning, and the ability to improve mental rotation by training is important for education in a range of disciplines. Recent theories from both cognitive science and brain sciences are beginning to converge on the notion that improvements in mental rotation occur by learning to categorise, rather than rotate, stimuli. We will use complementary analyses of data from two brain imaging techniques (electroencephalogram - EEG - and magnetoencephalogram - MEG) to determine the brain changes that underlie improved mental rotation performance.

Aging and workload-capacity

John & Daphne Keats Endowment Grant, 2008-2009
Chief Investigator: A. Eidels
Associate Investigator: B.M. Ben-David

What happens to the efficiency of information processing as we increase the cognitive load, and how is it affected by aging? In the redundant-target design, an observer detects the presence of a target. A trial can include two (redundant), single, or no-targets. Comparing response latencies on single- vs. redundant-target trials yields a capacity coefficient function (Townsend & Nozawa, 1995), which in turn reflects on the efficiency of integrating different sources of information. I plan to explore whether older-adults integrate visual and auditory stimuli (with and without distractors) differently than younger adults. We plan to test various theories that were put forth to account for informationprocessing slowdown in older adults: Generalized cognitive slowing models with a single age-related slowing equation; Information degradation models linking sensory loss with cognitive tasks; and models assuming a decrease in the efficiency of inhibiting distractors. Finally, we plan to develop processing model(s) that accommodate the data.

An integrated approach to absolute identification

John & Daphne Keats Endowment Grant, 2008-2009
Chief Investigators: P. Dodds, S.D. Brown & A.Heathcote

In a typical absolute identification (AI) task, participants engage in a form of stimulus magnitude estimation: they are presented with a series of stimuli that vary on only one dimension, each labelled with a unique label. They are then presented with these stimuli one at a time, and asked to try and remember the label that was previously associated with it. For over 50 years, performance has been assumed to be very poor, even after significant practice (e.g. Garner, 1953). Recent work by Rouder, Morey, Cowan and Pfaltz (2004) however, has shown that people are able to significantly increase their performance in at least one form of absolute identification task. We have successfully replicated this result and now aim to further explore the limits of practice, with a focus on whether this learning effect is apparent for other stimulus modalities, as well as determine what is that people are learning.

Accounting for old item variability in recognition memory for words

John & Daphne Keats Endowment Grant
Chief Investigators: M. Prince & A.Heathcote, 2008-2009

One of the most widely used models for drawing conclusions about recognition memory is signal detection theory, which represents targets (i.e., old items) and lures (i.e., new items) as two Gaussian distributions located on a single strength dimension. The strongest support for an unequal variance signal detection model is offered by receiver operating characteristics (ROCs), which plot the hit rate for old items against the false alarm rate for new items across a range of decision criteria. ROC analyses further show that old items are more variable than new items. However, the cause of this higher variability has not been thoroughly examined. The current research aims to determine what makes old items more variable. To achieve this we will use both a cutting edge linear mixed effects package (lme4, Baayen, Davidson & Bates, in press) and Bayesian analyses (e.g., Morey, Pratte, & Rouder, in press) that allow subjects and items to be included as crossed additive item and subject random effects.

The time course of forgetting

John & Daphne Keats Endowment Grant, 2007-2008
Chief Investigators: L.A. Averell & A.Heathcote

The proposed project will investigate quantitative models of forgetting with particular reference to Jost's second law of retention as well as the ongoing debate regarding the utility of above chance asymptotic retention in both explicit and implicit memory tasks. In a recent series of high impact publications Wixted (2004a, 2004b, 2005, in press) concluded that behavioural data agrees with at least two of the main indicators of Jost's law, namely, that (a) forgetting curves are best described by a power function and (b) forgetting curves decay to zero over time. However, Wixted underestimated the distortion caused by averaging non-linear data and the effects of model mimicry and functional complexity. The proposed experiments will attend to these problems in the extent literature with tighter control over longer time frames than has being previously attempted.

Testing a truism: people cannot learn absolute identification.

John & Daphne Keats Endowment Grant, 2007-2008
Chief Investigators: C. Donkin, S.D. Brown & A.Heathcote

Absolute identification (AI) is the task of attaching a label (such as #1, #2, . ,#10) to each item in a unidemensional set (such as a set of ten lines of increasing length). AI has fascinated researchers for 50+ years, because of the counter-intuitive phenomena that are regularly observed. One such phenomenon is that performance in an AI task quickly reaches asymptote, after which increased practise provides no further improvement. Rouder, Morey, Cowan and Pfaltz (2004) recently challenged this well-known fact, finding large and sustained learning effects in an AI task. We have replicated Rouder et al.'s findings and, in a follow-up experiment, found evidence contradicting their account of the data. We propose a series of experiments to discover the underlying factors that facilitate the unusual learning effect.

Integrating mathematical models of human cognition with neurophysiological data from fMRI.

Australian Academy of Sciences Scientific Visits Grant
Chief Investigators: Scott Brown & E.-J. Wagenmakers (UvAmsterdam)

In this project, we will use data from brain imaging experiments undertaken at UvAmsterdam and the Max Planck Institute to inform our continuing development of mathematical theories of human memory and decision making such as the linear ballistic accumulator model and the E-Z diffusion model.