This thesis explores contributions of prefrontal cortex (PFC) to memory using positron emission tomography (PET) and functional magnetic resonance imaging (fMRI). I begin by considering the cognitive neuroscience of memory processes and the impact that functional neuroimaging may have upon this. I then describe a series of PET and fMRI experiments concerned, primarily, with dissociating frontal contributions to encoding and retrieval processes. These initial studies show that left PFC activation predominates at encoding and right PFC activation at retrieval. Four further studies of left prefrontal activation at encoding are presented. Together, they show that left PFC is sensitive to tasks that require the organisation of encoded material according to its semantic attributes and that a more dorsal region of lateral PFC may specifically reflect the requirement to select from amongst semantic attributes in order to meet specific demands of the tasks. This region, and the behavioural performance associated with it, is shown to be sensitive to interference produced both by competing semantic attributes and by a simultaneously performed, distracting motor task. The two experiments on memory retrieval that are presented here provide evidence for distinctive roles of right dorsolateral and ventrolateral PFC during retrieval of verbal material. The ventrolateral region appears to reflect the changing specification of search parameters that occurs at the outset of a memory search and the dorsolateral PFC activation pattern is consistent with a role in monitoring and verification processes optimising the retrieval process. In conclusion, I review the broader literature on neuroimaging of memory-related frontal cortical function. While there are a number of inconsistencies, I suggest that the results presented here fit into an emerging pattern indicating the importance of PFC in memory encoding and retrieval and the distinctive roles of dorsolateral and ventrolateral regions within and between these memory stages.

No description available.

The mediodorsal thalamus (MD) is implicated in recognition memory however its exact role is unclear. The aims of this study were to provide a detailed analysis of the role of MD in recognition memory in the rat using variants of the spontaneous object recognition task to assess item (object or odour) recognition, recency recognition or associative recognition memory separately. In the first series of experiments, bilateral lesions in MD or mPFC impaired both recency and associative recognition, but spared item recognition and object location performance. The second series of experiments used disconnection techniques to investigate whether an interaction between the MD and mPFC is necessary during these recognition memory processes. Unilateral lesions in MD and mPFC in contralateral hemispheres produced deficits in recency and object-in-place associative recognition but spared item recognition, and object-in-context memory. The final series of experiments investigated the neurochemical basis of object-in-place associative recognition memory. Intra-MD Infusions of either NBQX, a glutamatergic antagonist, or muscimol, a GABAergic agonist produced selective deficits in retrieval, but not in acquisition of object-in-place associative memory. In contract, intra-mPFC infusions of NBQX impaired both acquisition and retrieval, while muscimol had no effect. Interestingly, crossed infusions into both the MD and mPFC disrupted retrieval but not acquisition. These findings indicate that MD plays a selective role in recognition memory when associative or recency, but not single item, discriminations are made. Secondly these data show that to make these discriminations, MD must functionally interact with the mPFC. Finally, it was shown that during object-in- place associative recognition memory, MD and the interaction between MD and mPFC appears critical during the retrieval phase only. Thus together these results show for the first time that MD plays a selective role in the recognition memory via a thalamocortical interaction with mPFC, and that for associative recognition memory this interaction is required for the retrieval phase only.

Object-in-place (OIP) memory is a form of associative recognition memory which relies on the formation of an association between an object and the place in which such object was encountered. Existing evidence shows that OIP memory depends upon a circuit of neural regions which include the medial prefrontal cortex (mPFC), perirhinal cortex (PRH) and hippocampus (HPC). The aim of this thesis was to characterise further the cellular mechanisms which underlie OIP associative memory in the rat. Several lines of research have shown that mPFC, PRH and HPC receive a substantial dopaminergic projection and that in these areas dopamine plays a critical role in modulating plasticity and memory processes. In the first part of this thesis the role of dopamine neurotransmission within mPFC, PRH and HPC for OIP memory was investigated. Guide cannulae were implanted in the regions of interest to allow local infusion of the selective dopamine D1 receptor antagonist SCH23390 and OIP memory was tested using a spontaneous OlP memory task. Oopaminergic neurotransmission through D1 receptors was found to be required in mPFC for acquisition, but not for consolidation or retrieval, of both short-term and long-term OlP memory. In contrast direct infusion of SCH23390 into PRH or HPC produced no impairment in memory. Further, a possible role of dopamine through D2 receptors was excluded, as intra-mPFC infusion of the D2 receptor antagonist eticlopride did not cause any impairment in the task. These results demonstrated that dopaminergic neurotransmission via 01 receptors was selectively involved in the acquisition of OIP memory in the mPFC. The next series of experiments investigated the synaptic plasticity mechanism underlying long-term (24 h) OIP memory within the mPFC. Local blockade of PKMzeta, a protein critical for LTP, through infusion of the selective inhibitor ZIP, impaired OIP memory performance either when infused prior the sample or 5 h before the test. an the contrary, infusion of the synthetic peptide GluR2-3Y, a molecule critical for LTO, either before the sample phase or 5 h before the test, did not show any memory impairment. These findings suggested that an LTP-like, rather than an LTO-like, mechanism occurs both during the encoding and the maintenance of OIP memory. The final series of experiments presented in this thesis aimed to investigate whether other brain regions, beyond mPFC, PRH or HPC, were involved in OlP associative recognition memory. Neuronal activity was measured through the differential expression of the immediate early genes c-fos and zif268, following presentation of novel or familiar arrangements of visual stimuli. However here were no significant differences in levels of neuronal activation following the presentation of novel and familiar stimuli in any of the areas analysed, a finding which may relate to procedural issues in the presentation of the visual stimuli.

Rhythmic network oscillations are observed in cortical structures during many cognitive states in vivo including memory formation, processing and consolidation and are implicated in the temporal coding of sensory information. The basolateral amygdala (BLA) has a fundamental role in affective memory processing.

This thesis reviews the evidence for dual process models of recognition memory, where both recollection and familiarity are thought to support overall performance. In particular, the focus is on one computational interpretation; the Complementary Learning Systems (CLS) model (Norman & O'Reilly, 2003). This is a biologically constrained dual process model based on the neuroanatomy and function of regions within the medial temporal lobes.

No description available.

Neuroimaging literature has identified several regions involved in encoding and recognition processes. A review of the literature illustrated considerable variations in the precise location and mechanisms of these processes, and it was these variations that were investigated in the studies in this thesis. Magnetoencephalography (MEG) was used as the neuroimaging tool and a preliminary study identified Synthetic Aperture Magnetometry (SAM) and not a traditional dipole fitting technique, as an appropriate tool for identifying the multiple cortical regions involved in recognition memory. It has been suggested that there is hemispheric asymmetry in encoding and recognition processes. There are two main hypotheses: the first suggesting that there is task-specificity, the second that this specificity is determined by stimulus modality. A series of experiments was completed with two main aims: first to produce consistent and complementary recognition memory data with MEG, and second to determine whether there exists any hemispheric asymmetry in recognition memory. The results obtained from five experiments demonstrated activation of prefrontal and middle temporal structures, which were consistent with those reported in previous neuroimaging studies. It was suggested that this diverse activation may be explained by the involvement of a semantic network during recognition memory processes. In support of this, a subsequent study involving a semantic encoding task demonstrated that category-specific differences in cortical activation also existed in the recognition memory phase. Controlling for the involvement of such semantic processes produced predominantly bilateral activation. It was suggested that the apparent hemispheric asymmetry findings reported in the literature may be due to the 'coarse' temporal analysis available with earlier imaging techniques, which over-simplified the networks reported by being unable to recognise the early complex processes associated with semantic processing which these MEG studies were able to identify. The importance of frequency-specific activations, specifically theta synchronisation and alpha desynchronisation, in memory processes was also investigated.

thesis submitted in partial fulfilment of the requirements for the degree of Doctor of Philosophy at the University of Oxford Approximate Word Count: 50,000 After literary scholars, the group of people who most likely cite Marcel Proust are the niche group of psychologists and neuroscientists researching the topic of memory. The incident of the madeleine and Proust's vibrant re-experiencing of 'times past' highlights how important contextual associations are in our lives. The memories we form are often rich in contextual detail, and it is this type of memory which I aim to explore in this thesis. Specifically, I show how memories of contextual nature are formed, and used to guide behaviour. In the General Introduction (Chapter 1), I review the background literature of attention, and the different sources of information that guide it, as well as how contextual information -the associations between iterns-, specifically in natural scenes, can serve as such a source. Next, I describe in detail the literature to date on memory-based signals for attentional guidance. The next chapter summarizes the methodological approaches used in this thesis (Chapter 2). In Chapter 3, I show that long-term memory can optimize perception in complex natural scenes by modulating preparatory attention as well as target processing, using electroencephalography (EEG). In Chapter 4, exploiting the high temporal and spatial resolution of magnetoencephalography (MEG), I explore the neurophysiological markers of encoding, while participants learned contextual associations. In the final experimental chapter (Chapter 5), in a series of experiments I test the low-level mechanisms through which the long-term memory-bias in attentional guidance comes about. In the General Discussion (Chapter 6), I summarize my Findings and incorporate them into the existing literature, and propose outstanding questions.