Earned Memory

OF THE DISSERTATION Application-Specific Memory Subsystems by Joseph G. Wingbermuehle Doctor of Philosophy in Computer Science Washington University in St. Louis, 2015 Professor Roger D. Chamberlain, Chair The disparity in performance between processors and main memories has led computer architects to incorporate large cache hierarchies in modern computers. These cache hierarchies are designed to be general-purpose in that they strive to provide the best possible performance across a wide range of applications. However, such a memory subsystem does not necessarily provide the best possible performance for a particular application. Although general-purpose memory subsystems are desirable when the work-load is unknown and the memory subsystem must remain fixed, when this is not the case a custom memory subsystem may be beneficial. For example, in an application-specific integrated circuit (ASIC) or a field-programmable gate array (FPGA) designed to run a particular application, a custom memory subsystem optimized for that application would be desirable. In addition, when there are tunable parameters in the memory subsystem, it may make sense to change these parameters depending on the application being run. Such a situation arises today with FPGAs and, to a lesser extent, GPUs, and it is plausible that general-purpose computers will begin to support greater flexibility in the memory subsystem in the future. In this dissertation, we first show that it is possible to create application-specific memory subsystems that provide much better performance than a general-purpose memory subsystem. In addition, we show a way to discover such memory subsystems automatically using a superoptimization technique on memory address traces gathered from applications. This allows one to generate a custom memory subsystem with little effort. xi We next show that our memory subsystem superoptimization technique can be used to optimize for objectives other than performance. As an example, we show that it is possible to reduce the number of writes to the main memory, which can be useful for main memories with limited write durability, such as flash or Phase-Change Memory (PCM). Finally, we show how to superoptimize memory subsystems for streaming applications, which are a class of parallel applications. In particular, we show that, through the use of ScalaPipe, we can author and deploy streaming applications targeting FPGAs with superoptimized memory subsystems. ScalaPipe is a domain-specific language (DSL) embedded in the Scala programming language for generating streaming applications that can be implemented on CPUs and FPGAs. Using the ScalaPipe implementation, we are able to demonstrate actual performance improvements using the superoptimized memory subsystem with applications implemented in hardware.

<b>Purpose: </b>The study aimed to test a combination of semantic memory and traditional episodic memory therapies on episodic memory deficits in adults with traumatic brain injury.<b>Method: </b>Twenty-five participants who had been diagnosed with traumatic brain injury and had episodic memory deficits were randomly assigned either to a combined memory treatment group (<i>n</i> = 16) or to a wait-list control group (<i>n</i> = 9). Before and after treatment, they completed standardized neuropsychological testing for episodic memory and related cognitive domains, including the California Verbal Learning Test–Second Edition, the Controlled Oral Word Association Test, the University of Southern California Repeatable Episodic Memory Test, the Wechsler Abbreviated Scale of Intelligence–Second Edition Matrices, the Test of Everyday Attention, the Memory Assessment Clinics Self-Rating Scale, the Expressive Vocabulary Test–Second Edition, and the Story Recall subtest from the Rivermead Behavioural Memory Test. In addition to a traditional episodic memory therapy, the treatment group received a novel semantic memory–focused therapy, which involved participants finding meaningful connections between diverse concepts represented by sets of two or three words.<b>Results:</b> The treatment group demonstrated statistically significant improvement in memory for list learning tasks, and there was a significant difference from pretest to posttest between the treatment group and the wait-list control group. Clinical significance was demonstrated for the treatment group using minimally important difference calculations.<b>Conclusion: </b>Combined memory therapy resulted in significant improvements in episodic memory, semantic memory, and attention, in comparison to no treatment.<b><br></b><b>Supplemental Material S1. </b>Time-ordered session treatment plan.<br>D'Angelo, E. C., Ober, B. A., &amp; Shenaut, G. K. (2021). Combined memory training: An approach for episodic memory deficits in traumatic brain injury.<i> American Journal of Speech-Language Pathology</i>. Advance online publication. https://doi.org/10.1044/2020_AJSLP-20-00075<br><b>Publisher Note: </b>This article is part of the Special Issue: Select Papers From the International Cognitive-Communication Disorders Conference.<br>

Distributed shared memory (DSM) systems allow programmers to use a shared memory programming model on distributed systems. However, DSM systems present new challenges to performance evaluation and debugging. While programmers write programs with a shared memory model using shared variables and shared memory synchronization primitives to coordinate processes, most performance problems arise from the implementation of these abstractions with messages. Because the DSM abstraction hides the underlying message passing, it is difficult to identify the sources of performance problems. We have developed an instrumentation system and performance evaluation techniques for DSM systems that correlate message passing behavior in the lower levels of a DSM system with the higher-level shared memory programming abstractions. We have developed a prototype tool called Xunify to demonstrate the feasibility of such an approach. Xunify provides graphical animations of DSM programs at three distinct levels and offers a powerful event filtering facility capable to help show system behavior corresponding to specific DSM constructs.

A wide body of animal and human research suggests that MDMA/ecstasy use is associated with lasting serotonergic neurotoxicity, particularly in mediotemporal and frontostriatal brain regions (Benningfield, 2013; Green, Mechan, Elliott, O'Shea, & Colado, 2003). Serotonin is important for a wide range of neurocognitive and psychobiological processes, and therefore neurotoxicity in these regions may have significant functional implications for ecstasy users. The objective of this thesis was to examine the neurocognitive and psychobiological effects of ecstasy that may be associated with serotonergic damage to mediotemporal and frontostriatal regions. Specifically, we examined this using an overnight memory consolidation paradigm. Overnight memory consolidation refers to the process of memory enhancement and increased resistance to interference that occurs across an episode of sleep, relative to a similar time-frame of wakefulness (Stickgold & Walker, 2013). The consolidation of declarative material in healthy participants is associated with prefrontal slow wave activity normally observed during deep sleep (Mander et al., 2013). The findings outlined above of serotonergic dysfunction in frontostriatal regions, coupled with evidence that ecstasy users have disrupted sleep architecture (Ogeil, Rajaratnam, & Broadbear, 2012) led us to hypothesise that ecstasy users will show impairments in the overnight consolidation of memory. We examined this hypothesis in two overnight consolidation studies. The first explored the overnight consolidation of emotional material using a visual recognition task delivered in an online environment. We found that compared to drug-naive controls (N=16), ecstasy users (N=15) tested after a period of sleep had stronger memory for negative visual images relative to neutral visual images. We suggest that this is due to a negative attribution bias, whereby material with negative content is preferentially consolidated and better recalled relative to material with neutral or positive content. The second overnight consolidation study utilised a verbal memory task to explore overnight memory consolidation in a laboratory setting. The task assessed whether sleep protects against interference from similar material, a phenomenon that has been confirmed in several studies among drug-naive controls (e.g. Ellenbogen, Hulbert, Jiang, & Stickgold, 2009). The results indicated that ecstasy users (N=12) did not show the same resistance to interference that was conferred to control participants (N=13) by overnight consolidation processes. Furthermore, our objective measurement of sleep architecture indicated that ecstasy users spent less time in deep sleep relative to controls, indicating that less time was spent in sleep stages that are known to be important for overnight memory consolidation. Together, this data suggests that ecstasy users experience dysfunctional sleep leading to impaired overnight memory consolidation. Finally, additional functional implications of disrupted mediotemporal and frontostriatal circuitry were examined using neurocognitive tasks that recruit these regions. Consistent with past research, we found reduced performance in a (same day) declarative long term memory task in ecstasy users. We also found that ecstasy users were selectively impaired on tasks of encoding, proactive interference memory, updating of working memory and complex planning, and therefore propose that executive functioning tasks engaging a higher cognitive load might reveal more subtle neurocognitive impairments experienced by ecstasy users. This thesis highlights the presence of selective neurocognitive executive dysfunction among ecstasy users. Furthermore, this thesis provides a unique contribution to the literature by demonstrating the key role of disrupted overnight consolidation processes in subsequent dysfunctional memory recall.

The Novel Encounter: Visualising Memories of Home is a personal examination of both the absence and presence of home. I explore the concept of home through the lens of memory as well as a representation of mundane space. I approach these concepts from the perspective that home is not necessarily a fixed or ideal place, but rather an on-going pursuit underpinned by the tension between preservation and transformation. Within these parameters, I define home as a self-referential process whereby memory and autobiography are integral. Home and memory are reconceptualised through visual language as a hybrid of image and narrative. Central to my research is how visual communicators can engage with and articulate home and memory as a portable artefact and a core subject for storytelling, which embraces ambiguity and the potential for loss through the abstract and poetic language of the photogram and the fragility of paper. Building on the work of philosophers and theorists, I examine how psychological, phenomenological, autobiographical, and practical contributions about home and memory inform my studio investigation. Artists and designers who have framed home and memory as a subject, with a particular focus on empty interiors, absence and the repetitiveness of the ordinary, are explored. I outline where these concepts and practices intersect with the language of visual communication and examine how an autobiographical investigation of home through memory can enter into this domain. To address these questions, my studio work aims through a combination of photograms, cut-paper typography, illustrations, and paper sculpture to develop a method of recording, cataloguing and reconceptualising memory set within the rooms where personal life is lived. My research seeks to contribute to new knowledge in the practice and education of visual communication, which combines visualisation, narrative, experiential and material aspects of domestic space with the ephemeral qualities of memory. I argue that home, memory and autobiography as a genre of visual communication support opportunities for authorship and in doing so it enrich methodologies of storytelling.

Memory for Time Heather Van Volkinburg Research on the perception of time has focused on isolating an internal time keeping mechanism. This focus has caused research in this domain to stay focused on perceptual mechanisms. However, information that has been perceived must also be stored into memory. The dominant model of time perception, SET, specifies a clock stage, a memory stage, and a decision stage (Gibbon, 1977), but there has been comparatively more research focusing on the internal clock than on memory mechanisms. This dissertation focuses on the memory for time by incorporating manipulations used in non-temporal memory research into timing tasks. Chapter 1 targets working memory for time and shows that brief delays between learning and recall cause temporal estimates to lengthen. Chapter 2 targets retention and storage of multiple time intervals over a delay of a few minutes and shows that the estimates of target times learned together will migrate towards each other in memory. We also showed that scalar variability arises at retrieval. Chapter 3 attempts to reconfigure a categorization task to target implicit memory for time. Overall, the research demonstrates that exploring memory mechanisms for time will increase our understanding of time perception and provide us with information that focusing on the internal

Long-range synchrony between medial prefrontal cortex, thalamus and hippocampus underlies working memory behavior in mice. Pia-Kelsey O’Neill Presently, there are no antipsychotic drugs capable of treating the cognitive dysfunctions of schizophrenia. In order to inform the development of better therapies, it is essential to understand the mechanism behind dysfunctional cognition, which requires an understanding of functional cognition. Spatial working memory, a measure of cognitive function, can be assessed in the mouse using a task of delayed alternation: the T-maze. In this thesis, I focus on spatial working memory behavior in the mouse and three brain regions that are implicated in this behavior: the medial prefrontal cortex (mPFC), the hippocampus (HPC) and the medial dorsal thalamus (MD). Lesion and electrophysiological studies in each structure have demonstrated their importance during working memory behavior. Disconnection studies also show that the coordination between the mPFC and either the HPC or MD is important for the behavior, but little is known about the mechanism by which they coordinate. The MD and the ventral region of the hippocampus (vHPC) have robust projections into the mPFC. They are therefore in a good position to influence mPFC activity. Previous reports show that the mPFC and the dorsal region of the hippocampus (dHPC) synchronize activity in the theta range (4-12 Hz) with working memory demand. However, the dHPC does not directly connect with the mPFC so it is unclear how this coordination occurs. We hypothesized that the vHPC may also be involved in spatial working memory behavior and that it may mediate the dHPC-mPFC theta synchrony observed. To test these hypotheses, we recorded neural activity simultaneously from the mPFC, dHPC and vHPC in mice performing the T-maze task. Local field potential oscillations (LFPs), thought to be a measure of synchronized synaptic activity, were obtained from each area. We observed an increase in theta synchrony between the mPFC and both the dHPC and vHPC. Removing the influence of vHPC both analytically and experimentally, we found a decrease in synchrony of the dHPC-mPFC. Aside from the disconnection studies, little is known about the MD-mPFC pathway in rodents. However, due to evidence from schizophrenia patients of altered correlation specifically between the MD and PFC, we hypothesized that an electrophysiological correlate of working memory exists in the MD-mPFC pathway as well and that a decrease in MD activity may lead to prefrontal dysfunction. To test these hypotheses, we recorded LFPs from the mPFC and both single unit activity and LFPs from the MD in mice performing the T-maze task. We observed an increase in phase locking of MD cells to mPFC LFPs in beta (13-30Hz) range during the choice phase of the task. We then utilized a pharmacogenetic technique to decrease firing rate in a small portion of MD cells, which resulted in a deficit in both task acquisition and performance. The increase in MD-mPFC beta phase locking we had observed was not present in MDinactivated animals. Interestingly, beta coherence between the two structures across learning was highly correlated with choice accuracy on the task. This suggests that MD-PFC coordination is predictive of working memory performance. These findings illustrate how long-range synchrony of the mPFC with HPC in the theta frequency range and with the MD in the beta frequency range may be important markers for normal working memory behavior and if disrupted in humans, could contribute to the cognitive symptoms of schizophrenia.

Background: Memory impairment is a well-known effect of many benzodiazepine compounds which is mediated through their action on gamma-aminobutyric acid type A (GABAA) receptors. On the other hand, cannabinoids can affect learning and memory process through presynaptic modulation of the release of both excitatory glutamate and inhibitory GABA transmitters in brain regions involved in learning and memory. The aim of the present study was to investigate the effect of cannabinoids on memory impairment and long-term potentiation (LTP) reduction properties of the short acting benzodiazepine midazolam.&lt;br /&gt;Materials and Methods: One week after insertion of guide cannula by stereotaxic surgery, cannabinoid compounds or midazolam were administered by intracerebroventricular (i.c.v.) injection into lateral ventricle of male rats. Spatial memory task was evaluated using Morris water maze (MWM) test. Electrophysiological evaluation was done by field potential recording of hippocampal neurons in unconscious rats.&lt;br /&gt;Results: In MWM test, while i.c.v. administration of AM251 (200 and 500 ng) per se could not change learning and memory function in rats, pretreatment of rats with AM251 (500 ng; i.c.v.) attenuated midazolam-induced memory impairment. In field potential recording, while i.c.v. administration of AM251 (500 ng) and WIN55212-2 (10 μg) did not have any effect on population spike amplitude, pretreatment of rats with both AM251 and WIN55212-2 significantly diminished midazolam-induced PS amplitude reduction in hippocampal neurons.&lt;br /&gt;Conclusion: Our&lt;br /&gt;Our results suggest the involvement of cannabinoid CB1 receptors in both memory impairment and LTP reduction in hippocampal neurons which was produced by midazolam. This interaction is likely through their effect on both GABAergic and glutamatergic receptors in hippocampus.

The paper starts from a qualitative research, carried out through 50 in-depth interviews, on memory and its evolutions in connection with digital tools and networked environments. The aim was to investigate how new generations build their individual memory and more specifically which things, mementos and interfaces of memory they use; which are their practices to create and to socialize memory, how and in which measure they transform it in cultural capital. The role of interactive media for memory production, storage and communication will be outlined; furthermore, the web and social media will be analysed as cultural interfaces through which people approach their past. The analysis will explore how digital environments for writing, publishing, showing personal experience are changing the subjective criteria to distinguish past and present, private and public dimensions of memory. The productive dimension of 2.0 web environments is indeed giving result to a new diffused subjectivity in which, on the one hand people are stimulated to enlarge the virtual audience of their subjective memories, whilst on the other hand, they transform and mediate their subjectivity in order to fit it in a global language and circuit.

Solving the Credit Assignment Problem: The interaction of Explicit and Implicit learning with Internal and External State Information Wai-Tat Fu (wfu@cmu.edu) Human Factors Division and Beckman Institute University of Illinois at Urbana-Champaign 1 Airport Road, Savoy, IL 61874, USA John R. Anderson (ja+@cmu.edu) Department of Psychology Carnegie Mellon University 5000 Forbes Avenue Pittsburgh, PA 15213 USA recognition of external state information (signs on the walls). Indeed, many have argued that real-world skills often involve the interplay between cognition (internal), perception, and action (external) that the understanding of these interactive skills requires careful study of how internal (memory) and external information (cues in the environment) are processed in the learning processes (Ballard, 1997; Fu & Gray, 2000; 2004; Gray & Fu, 2004; Larkin, 1989; Gray, Sims, Fu, & Schoelles, in press). The navigation problem above is an example of one of the most difficult situations in skill learning: when the learner has to perform a sequence of actions but only gets feedback on their success at the end of the sequence (e.g., when the destination is reached). This creates a credit-assignment problem, in which the learner has to assign credits to earlier actions that are responsible for eventual success. When actions are interdependent, either memory of previous actions or recognition of the correct problem state in the external environment is required to properly assign credits to the appropriate actions. In this article, we present results from an experiment in which we study how people learn to solve the credit-assignment problem in a simple but challenging example of such a situation. Our focus is on the recent proposal that humans exhibit two distinct learning processes and we apply it to learning of action sequences with delayed feedback: an explicit process (with awareness) that requires memory for actions and outcomes, and an implicit process (without awareness) that does not require such memory. We will first review research in some related areas that informed the design of our experiment. Abstract In most problem-solving activities, feedback is received at the end of an action sequence. This creates a credit-assignment problem where the learner must associate the feedback with earlier actions, and the interdependencies of actions require the learner to either remember past choices of actions (internal state information) or rely on external cues in the environment (external state information) to select the right actions. We investigated the nature of explicit and implicit learning processes in the credit-assignment problem using a probabilistic sequential choice task with and without external state information. We found that when explicit memory encoding was dominant, subjects were faster to select the better option in their first choices than in the last choices; when implicit reinforcement learning was dominant subjects were faster to select the better option in their last choices than in their first choices. However, implicit reinforcement learning was only successful when distinct external state information was available. The results suggest the nature of learning in credit assignment: an explicit memory encoding process that keeps track of internal state information and a reinforcement-learning process that uses state information to propagate reinforcement backwards to previous choices. However, the implicit reinforcement learning process is effective only when the valences can be attributed to the appropriate states in the system – either internally generated states in the cognitive system or externally presented stimuli in the environment. Introduction Consider a person navigating in a large office building. The person has to decide when to turn left or right at various hallway intersections. The sequence of decisions is interdependent – e.g., turning left at a particular hallway intersection will affect the decisions at the next intersections. The person may therefore need to keep track of previous actions to inform what actions to take in the future. In reality, memory of previous actions (internal state information) may not be necessary as people can explicitly seek information in the environment (external state information) to know where one is located or which direction to go to reach a destination (Fu & Gray, 2006). Learning to navigate is therefore likely to involve both the retention of internal state information (memory) and the Explicit and Implicit Learning Probability Learning and Classification There have been numerous studies on the learning of the probabilistic relationship between choices and their consequences. The simplest situation is the probability- learning experiment in which subjects guess which of the alternatives occurs and then receives feedback on their guesses (e.g., Estes, 1964). One robust finding is that subjects often “probability match”; that is, they will choose a particular alternative with the same probability that it is reinforced (e.g., Friedman et al., 1964). This leads many to propose that probability matching is the result of an implicit

Abstract. The following issues regarding to organization memory were mainly discussed in extant literature: 1. What is organization memory – organizations as knowledge bases. 2. Where is the memory stored? This article offers a compact commentary for these two questions, revisiting from a where-what perspective – that we should consider the nature of the context where known memory is embedded then we could really know the nature of what that knowledge piece is. Implications for research were elaborated. Keywords. Organization memory, Knowledge bases, Known memory. JEL. D80, L22, L23.

In this thesis, the JavaDL logic is amended by a new modality for the verification of termination-insensitive, object-sensitive, flow-sensitive noninterference of memory locations in object-oriented programs, that is local variables and heap object fields. Noninterference is a security model applied in information flow analysis and the amended JavaDL logic improves the information flow analysis of Java Card programs using the JavaDL sequent calculus. JavaDL formulae are a superset of first-order logic formulae and the logic additionally contains dynamic logic modalities as well as a concept called updates for the transition and evaluation of Java Card program states. Noninterference of memory locations is a property of programs and the respective pairs of start and end states of their executions. The satisfaction of the property implies that a particular set of memory locations does not interfere with the memory locations not included in this set. A set of memory locations does not interfere with another set if the evaluations of the included memory locations in the end state of any execution are independent of the evaluations of the memory locations included in the other set. Noninterference in Java Card programs has already been formulated and verified in JavaDL by a concept called self-composition which compares the end states of two independent executions of the same program with respect to a particular set of memory locations. This investigation is realisable with the means provided by JavaDL, in particular the dynamic logic modalities. However, in many cases the double execution poses unnecessary proof overhead and the new modality defined in this thesis aims at reducing this overhead. The new calculus reaches this goal by partially interpreting the verified program on a single execution path. As soon as this execution path branches the calculus must return to self-composition or the analysis loses precision.

A software system can be called a safe-system with respect to memory, when it supports only strong-typing and it does not allow for the manual disposal of dynamic memory [2]. The first aspect guarantees that untyped, potentially dangerous operations are caught by the compiler or by run-time checks. The second issue is solved by the utilisation of an automatic memory reclamation scheme, i.e. a garbage collector.In this paper we argue that the careful choice of the programming language, along with an automatic memory reclamation scheme can optimise memory usage, while ensuring that many of the logical errors related to memory can be avoided.

The trend of using heterogeneous computing and HW/SW-Codesign approaches allows increasing performance significantly while reducing power consumption. One of the main challenges when combining multiple processing devices is the communication, as an inefficient communication configu-ration can pose a bottleneck to the overall system performance. To address this problem, we present a methodology that assists the designer in making good design decisions for systems using shared DDR memory for communication. Our methodology analyzes a software implementation of the application and subsequently predicts the memory accesses of a functionallyequivalent hardware implementation of the selected function. We furthermore propose an IP core that can perform these predicted memory accesses to estimate the achievable memory bandwidth between a functionally equivalent hardware implementation and shared memory. The resulting achievable memory bandwidth estimations differ by less than 2% from the actual achievable memory bandwidth of a functionally equivalent hardware implementation, demonstrating the feasibility of the presented methodology.

Foreword Acknowledgements A tribute to Edmond Roudnitska 1. Olfaction and cognition: a philosophical and psychoanalytical view Annick Le Guerer 2. Cognitive aspects of olfaction in perfumer practice Andre Holley 3. The specific characteristics of the sense of smell Egon Peter Koster 4. Names and categories for odors: the 'veridical label' Daniele Dubois 5. Nose-wise: olfactory metaphors in mind David Howes 6. Linguistic expressions of odors in French Sophie David 7. Classification of odors and structure-odor relationships Maurice Chastrette 8. The acquisition and activation of odor hedonics in everyday situations: conditioning and priming studies Dick Hermans and Frank Baeyens 9. Is there a hedonic dimension of odors? Catherine Rouby and Moustafa Bensafi 10. The influence of odors on mood and affective cognition Rachel S. Herz 11. Assessing putative human pheromones Suma Jacob, Bethanne Zelano, Davinder J. S. Hayreh and Martha K. McClintock 12. The neural correlates of emotion perception: from faces to taste Mary L. Phillips and Maike Heining 13. Testing odor memory: incidental vs. intentional learning implicit vs. explicit memory Sylvie Issanchou, Dominique Valentin, Claire Sulmont, Joachim Degel and Egon Peter Koster 14. Odor memory: a memory systems approach Maria Larsson 15. Repetition priming in odor memory Mats J. Olsson, Maria Faxbrink and Fredrik U. Jonsson 16. Odor memory in Alzheimer's disease Steven Nordin and Claire Murphy 17. Development of odor naming and odor memory from childhood to young adulthood Johannes Lehrner and Peter Walla 18. Odor coding at the periphery of the olfactory system Gilles Sicard 19. On human brain activity within the first second after odor presentation Bettina M. Pause 20. Processing of olfactory affective information: contribution of functional imaging studies Robert J. Zatorre 21. Experience-induced changes reveal functional dissociation within olfactory pathways Nadine Ravel, Anne-Marie Mouly, Pascal Chabaud and Remi Gervais 22. Increased taste sensitivity by familiarization to 'novel' stimuli Annick Faurion, Barbara Cerf, Anne-Maria Pillias and Nathalie Boireau 23. The cortical representation of taste and smell Edmund T. Rolls 24. New psychophysical insights in evaluating genetic variation in taste Katharine Fast, Valerie B. Duffy and Linda M. Bartoshuk 25. The individuality of odor perception Robyn Hudson and Hans DisTel 26. Olfactory cognition at the start of life: the perinatal shaping of selective odor responsiveness Benoist Schaal, Robert Soussignan and Luc Marlier 27. Age-related changes of chemosensory functions Thomas Hummel, S. Heilman and Claire Murphy Index.