Chimpanzees and bonobos exhibit divergent spatial memory development.

Spatial cognition and memory are critical cognitive skills underlying foraging behaviors for all primates. While the emergence of these skills has been the focus of much research on human children, little is known about ontogenetic patterns shaping spatial cognition in other species. Comparative developmental studies of nonhuman apes can illuminate which aspects of human spatial development are shared with other primates, versus which aspects are unique to our lineage. Here we present three studies examining spatial memory development in our closest living relatives, chimpanzees (Pan troglodytes) and bonobos (P. paniscus). We first compared memory in a naturalistic foraging task where apes had to recall the location of resources hidden in a large outdoor enclosure with a variety of landmarks (Studies 1 and 2). We then compared older apes using a matched memory choice paradigm (Study 3). We found that chimpanzees exhibited more accurate spatial memory than bonobos across contexts, supporting predictions from these species' different feeding ecologies. Furthermore, chimpanzees - but not bonobos - showed developmental improvements in spatial memory, indicating that bonobos exhibit cognitive paedomorphism (delays in developmental timing) in their spatial abilities relative to chimpanzees. Together, these results indicate that the development of spatial memory may differ even between closely related species. Moreover, changes in the spatial domain can emerge during nonhuman ape ontogeny, much like some changes seen in human children.

Cultural life scripts structure recall from autobiographical memory.

Three classes of evidence demonstrate the existence of life scripts, or culturally shared representations of the timing of major transitional life events. First, a reanalysis of earlier studies on age norms shows an increase in the number of transitional events between the ages of 15 and 30 years, and these events are associated with narrower age ranges and more positive emotion than events outside this period. Second, 1,485 Danes estimated how old hypothetical centenarians were when they had been happiest, saddest, most afraid, most in love, and had their most important and most traumatic experiences. Only the number of positive events showed an increase between the ages of 15 and 30 years. Third, undergraduates generated seven important events that were likely to occur in the life of a newborn. Pleasantness and whether events were expected to occur between the ages of 15 and 30 years predicted how frequently events were recorded. Life scripts provide an alternative explanation of the reminiscence bump. Emphasis is on culture, not individuals.

Memorabeatlia: a naturalistic study of long-term memory.

Seventy-six undergraduates were given the titles and first lines of Beatles' songs and asked to recall the songs. Seven hundred and four different undergraduates were cued with one line from each of 25 Beatles' songs and asked to recall the title. The probability of recalling a line was best predicted by the number of times a line was repeated in the song and how early the line first appeared in the song. The probability of cuing to the title was best predicted by whether the line shared words with the title. Although the subjects recalled only 21% of the lines, there were very few errors in recall, and the errors rarely violated the rhythmic, poetic, or thematic constraints of the songs. Acting together, these constraints can account for the near verbatim recall observed. Fourteen subjects, who transcribed one song, made fewer and different errors than the subjects who had recalled the song, indicating that the errors in recall were not primarily the result of errors in encoding.

Studying Recommender Systems to Enhance Distributed Computing Schedulers

Distributed Computing frameworks belong to a class of programming models that allow developers to

launch workloads on large clusters of machines. Due to the dramatic increase in the volume of

data gathered by ubiquitous computing devices, data analytic workloads have become a common

case among distributed computing applications, making Data Science an entire field of

Computer Science. We argue that Data Scientist's concern lays in three main components: a dataset,

a sequence of operations they wish to apply on this dataset, and some constraint they may have

related to their work (performances, QoS, budget, etc). However, it is actually extremely

difficult, without domain expertise, to perform data science. One need to select the right amount

and type of resources, pick up a framework, and configure it. Also, users are often running their

application in shared environments, ruled by schedulers expecting them to specify precisely their resource

needs. Inherent to the distributed and concurrent nature of the cited frameworks, monitoring and

profiling are hard, high dimensional problems that block users from making the right

configuration choices and determining the right amount of resources they need. Paradoxically, the

system is gathering a large amount of monitoring data at runtime, which remains unused.

In the ideal abstraction we envision for data scientists, the system is adaptive, able to exploit

monitoring data to learn about workloads, and process user requests into a tailored execution

context. In this work, we study different techniques that have been used to make steps toward

such system awareness, and explore a new way to do so by implementing machine learning

techniques to recommend a specific subset of system configurations for Apache Spark applications.

Furthermore, we present an in depth study of Apache Spark executors configuration, which highlight

the complexity in choosing the best one for a given workload.