Elena's Legacy. Texts written by Elena, excerpts from texts she liked, her artworks.

Wednesday, February 14, 2018

Neuroimaging Techniques

Spectacular discoveries in neuroscience have been rendered possible with advances in neuroimaging technology. Aside from other methodology, such as electrophysiology (single and multiple electrode recording, EEG – electroencephalography) and lesion studies, brain imaging has allowed neuroscientists to shed light on cortical mechanisms and processes. For example, by and large the most common method, fMRI (functional Magnetic Resonance Imaging), takes advantage of changes in blood flow and BOLD (Blood Oxygen Level Dependent) signals to localize activity in certain parts of the cerebrum.

Due to the cortex being active in its entirety (as opposed to extreme modularity outlined in discarded phrenological views), the subtraction method is employed to determine neural correlates. The two images are juxtaposed and subtracted, leaving pictures of only those clusters that were activated by given stimuli. Other popular methods include MRI (Magnetic Resonance Imaging), PET (Positron Emission Tomography) and CAT (Computerized Axial Tomography) scans, and TMS (Transcranial Magnetic Stimulation).

Psychology & Statistics

The purpose of this essay is briefly explain the importance of understanding statistics in the psychological field. Psychologists rely on experiments and other research to gather evidence and construct theories. Generally, after formulating a hypothesis a researcher (or a group of researchers) will design an experiment to test the prediction or educated guess. Likewise, a review of the literature published on the subject is usually undertaken to get a broader picture. Also, articles which do not cover any primary research but review existent research on a subject, are called ‘review articles’ and a meta-analysis may or may not have been carried on the findings reviewed. The type of measures used also depends on the field one is most interested in psychology; to illustrate, neuroscientists may prefer to turn to fMRI (functional Magnetic Resonance Imagery) studies and neuron staining, while psychopathologists might first look at lesion studies and clinical cases.

An integral part of psychological theories comes from questioning, such as surveys and questionnaires. Indeed, most psychology (and business) students had to design at least one questionnaire during their undergraduate or college years. Questionnaire and interview questions may be open-ended or closed-ended. The two different types of questions correspond to qualitative and quantitative answers, respectively. An example of an open-ended question would be: How did you spend last summer? Conversely, closed-ended questions may be forced choice (such as answering yes or no, or fill in the blank), and multiple choice. Multiple choice questions are usually coded on the 5-point Likert-type scale and the results are then analysed using statistical methods.

Office Furniture. Illustration: Megan Jorgensen (Elena)

Psychologists also use other types of research methods, such as naturalistic observation, which involves observing an activity without intervening in any way. Precisely, the Hawthorne effect refers to the fact that people may behave differently when aware of being observed (or participating in an experiment). Likewise, psychologists often draw conclusions from correlations, but without implying causality. Thus, a correlation simply means that two variables move together, a positive correlation means that as X increases/decreases, Y also increases/decreases; while a negative correlation shows that the two variables move in opposite directions (as X increases/decreases, Y decreases/increases). Further, to meet the validity criterion, experimental results must be reliable and replicable.

Furthermore, longitudinal studies are carried over long periods of time, sometimes across generations. Alternatively, genetic studies may focus on fraternal or identical (monozygotic) twins. For example, studies on identical twins have shown that schizophrenia onset appears to follow a stress-diathesis model, in which having the genes is sufficient to develop the mental illness solely in the presence of environmental stressors (as evidenced by cases when in an identical twin pair, both of whom share the same genetic make-up, but only one of whom develops schizophrenia). Although the concordance and likelihood remain very high the closer the relative.

Perhaps one of the most fantastic anecdotes about genetic twin studies is the study of two identical twins who were separated at birth and adopted into two different families, in different Commonwealth countries. However, when the two brothers reunited in their forties, they noticed that not only had they chosen the same profession, led the same lifestyle and even married, divorced and remarried women with the same names, they even wore the same clothes. Such an argument lends credence to the nature side of the nature-nurture heated debate. Nowadays, most psychologists would agree that an interplay, or interaction, of both genes and environment, shapes individuals’ physical and mental health, personalities and other traits. Thus, the purpose of the present paper was to highlight the fundamental role that statistics and statistical analysis play in understanding psychology.

Computational Neuroscience

If I cannot build it, I do not understand it - Richard Feynman, Nobel prize in physics.

The present essay describes computational neuroscience as it applies to psychology. Indeed, the biology of the brain and the nervous system in general, or neurobiology, is a fascinating field. So is the related and very similar discipline of neuroscience. Evidently, with technological advances, computers began to take place in the elucidation of these highly complex matters. Therefore, the aim of the present section is a brief overview of the derivative sub-discipline of computational neuroscience. Computational neuroscientists build artificial models that attempt to mimic brain functions. Sounds like artificial intelligence? Maybe, but, arguably, science is still only working towards that goal. Although, of course, it all depends on the definition one accords to intelligence. After all, today’s machines can read, write, speak, count and perform many other complicated tasks necessitating processes akin to cognition and other mental feats in human.

The biology of the brain and the nervous system in general, or neurobiology, is a fascinating field. So is the related and very similar discipline of neuroscience. Evidently, with technological advances, computers began to take place in the elucidation of these highly complex matters. Therefore, the aim of the present section is a brief overview of the derivative sub-discipline of computational neuroscience.

Interestingly, according to Granger (2001), the neuroscientific branch is about more than computers per se or their working models, but about constructing simulacra of cortical activity. Amazingly, the author suggests that in time these theories would even permit the generation of prosthetic brains! Robots – Artificially intelligent machines, with the unveiling of the first androids (human like robotic entities, unlike the above depiction) in Japan and South Korea, no longer the realm of science fiction alone.urther, the ideology relies on a set of principles such as that neural systems rely on stochastic differential equations (Carillo et al., 2011). Rolls (2011) places the historical beginning of the collection of theoretical constructs around the 1970s, with mathematician David Marr as one of the first contributors. Thus, the present paper briefly highlights some key factors descriptive of computational neuroscience.

Robots – Artificially intelligent machines, with the unveiling of the first androids (human like robotic entities, unlike the above depiction) in Japan and South Korea, no longer the realm of science fiction alone. Image: Copyright © Elena.

Computational neuroscientists build artificial models that attempt to mimic brain functions. Sounds like artificial intelligence? Maybe, but, arguably, science is still only working towards that goal. Although, of course, it all depends on the definition one accords to intelligence. After all, today’s machines can read, write, speak, count and perform many other complicated tasks necessitating processes akin to cognition and other mental feats in human.

Further, the ideology relies on a set of principles such as that neural systems rely on stochastic differential equations (Carillo et al., 2011). Rolls (2011) places the historical beginning of the collection of theoretical constructs around the 1970s, with mathematician David Marr as one of the first contributors.

References:

Carillo, J. A., Gonzalez, M. D. M., Gualdani, M. P. & Schonbek, M. E. (2011). Classical solutions for a nonlinear Fokker-Planck equation arising in computational neuroscience. ArXiv, September 6, 1-22.
Granger, R. (2011). How brains are built: Principles of computational neuroscience. Cerebrum, January: 1-17.
Rolls, E. T. (2011). David Marr’s vision: Floreat computational neuroscience. Brain: A Journal of Neurology, 134: 913-16.

Tuesday, February 13, 2018

Academic Attack 2: Beta Is a Fickle Short-Term Performer

Academic Attack 2: Beta Is a Fickle Short-Term Performer (and Sometimes It Fails to Work for a Full Decade)

The divergence of theory from evidence is even more striking in the short run: For some short periods, it may happen that risk and return are negatively related. In 1972, for example, which was an “up” market year, it turned out that safer )lower-beta) stocks went up more than the more volatile securities. Fortune magazine commented dryly on this well-publicized failure, “the results defied the textboos.” What happened was that in 1972 styles changed in Wall Street as institutional investors eschewed younger, more speculative companies, the “faded ladies” of the late 1960s, and became much more enamored of the highest-quality, most stable leading corporations in the so-called “first tier” of stocks. This was the Nifty Fifty craze. It became clear that beta could not be used to guarantee investors a predictable performance over a period of a few months or even a year.

Black, Jensen, and Scholes found a similar type of anomaly for the entire period from April 1957 through December 1965. Not only does the zero-beta return exceed the riskless rate, but during this period of nearly nine years, securities with higher risk produced lower returns than less-risky (lower-beta) securities. Substantial deviations from the relationship predicted by the CAPM were also found for many subperiods.

The experience of the 1980s provided even more dramatic evidence of the folly of relying on beta measures to predict realized rates of return. It turned out that for the entire decade of the 1980s realized mutual-fund returns bore no relationship to their beta measures of risk.

A fickle short-term performer. Photo by Elena

The following chart shows the relationship between mutual-fund returns during the 1980s and the beta measures of systematic risk. These are the same funds which displayed a positive relationship between realized returns and risk covering a much longer twenty-year time period. Note that there is no positive relationship between the beta risk measures and the mutual-fund returns (the correlation coefficient between betas and returns for the 1980s is essentially zero). Indeed, were it not for the one observation in the top right-hand corner of the graph, there would have been a tendency for high-beta portfolios to earn a lower rate of return (That fund in the top right-hand corner of the chart with the extraordinary record is the Magellan Fund).

Thus, investors who thought they could use the capital-asset pricing model to fashion higher-risk portfolios in order to achieve higher rates of return during the 1980s were sadly disappointed.

If we mention that beta summarizes the total systematic risk of securities, we must accept three uncomfortable conclusions: 1) In some short periods, investors may be penalized for taking on more risk; 2) in the long run, investors are not rewarded enough for high risk and are overcompensated fory buying securities with low risk; and 3) in all periods, some unsystematic risk is being valued by the market. Any of these results is a serious contradiction of the CAPM.

Sunday, February 11, 2018

The Audrey Hepburn

The Audrey Hepburn was called after a commander’s Daniels Rastoropnov’s fiancee, who patiently waited for him on Earth while he was exploring the asteroids belt.

When we found the sky of this tiny planet a kind of pinkish-yellow rather than the blue which had erroneously first been reported, the announcement was greeted by a chorus of good-natured boos from the public – people wanted the other planets to be, even in this respect, like the Earth.

And yet the Audrey’s landscapes are staggering, the vistas breathtaking.

A long television series was made about the planet. In fact, we all are oriented toward astronomy and we are engaged to it with our heart as well as the mind. Some television series aimed at popular audiences, visually and musically stunning, prove that the public is far more intelligent than it has generally been given credit for; the deepest scientific questions on the nature and origin of the world excite the interests and passions of enormous numbers of people.

(Extract from The Rain, the famous SF novel by Elena and George B.)

(Many illustrations in this book are based on the striking visuals prepared by NeuroscienceTv for the television series. In fact, books and television series have somewhat different audiences and admit different approaches. This novel and Neuroscience television series represent a hopeful experiment in communicating some of the ideas, methods and joys of science).