Today I was removed from the Indo-Trento Advanced Research Program as a Junior Research Fellow at the University of Hyderabad and told to find a new PhD advisor for pointing out mistakes in a research paper which my advisor, Prof Bapi Raju, was prodding me to use as the basis of my research. Upon asking what was wrong with the review I wrote, my advisor told me that he did not have sufficient time to explain what was wrong. Upon insisting more and explaining my position on using a principle or theorem via a formal logical system as a basis for a research statement rather than arbitrary guess work, his answer was that he was not well trained enough in logic to point out what was wrong with my statements but he still believed that I was wrong. He went on to say that I needed to adjust my attitude to be more “cognitively flexible” and not be stuck on the idea of using provable principles or physics while doing cognitive science research. He said that my lack of cognitive flexibility was the main reason he was removing me from the project and for me to find another advisor.
I don’t feel angry, I don’t feel sad. But I do feel very insulted. I also feel subverted as I was being repeatedly asked to compromise my ethics and to get on board with a research program that ignores basic principles like “information theory” which, has been proven to be domain independent and applicable to every conceivable system under all conceivable states of a system. Is a human, or any part of a human, not a system? Can there possibly be an empirical method that does not involve a system and is not directed towards the observation of a system by a system? Supposedly such questions are too philosophical and thereby need to be suppressed. And even if these questions are philosophical and contemplative, why is that an excuse to suppress the question? Right, even this question is philosophical and pertains to philosophy of philosophy in a recursive way so I suppose it ought to be suppressed more forcefully.
I was put down while being told that, “you have been saying this about information theory over and over again for the past three years but I haven’t seen any proof by you about it being applicable to systems as you claim, in all these years. You only believe it to be applicable. Do you even know about the ‘many worlds’ concept in philosophy? Have you even read about many worlds interpretation where there can be such worlds in which your claims need not be true. You do not read things deeply and are cognitively inflexible.” Actually, that is why I had stressed on starting with a precept that would be true under all interpretations and thus be “valid”.
The proof of information theory’s validity is in the paper titled “a mathematical theory of communication”. It provides a relationship between “information transfer” and the changes within the configuration of a system. If the thing being analyzed is not a system, then what is that thing? The thing being analyzed must necessarily be a system i.e. a collection of entities. Furthermore, how can there be any scope for cognition within a system without information transfer? Thus, for a system to be cognitive, it minimally needs to experience information transfer. How much more of a proof can anyone provide? And indeed I had already presented this topic during my doctoral review meeting in which other faculty were present. I had explained the process of defining a semantic space as a geometric manifold with concepts as points within the manifold where in, human rationale would be defined as the process of navigating from one point to next within the manifold. The constraints of this manifold that describe its nature as well as the possible ways of navigation from one concept to the next were delineated by me using information theory and using a definition of the “partionability” of the manifold. But these facts were not important, I was guided away from my original thesis to be used as an analyst for a research program that somehow mattered to my professor. I did go along trusting my advisor till I discovered how flawed the research he was prodding me towards was and at this juncture I could plainly see that my professor was fishing for a silly reason to dismiss me because I was not going to blindly go along with the program.
Perhaps the most offending thing I wrote about the particular research work I reviewed was how the authors of the paper arbitrarily focused on an anatomical region of the brain using an arbitrary size of that region (50 voxels) while ignoring the rest of the brain and that the authors had a bad excuse for eliminating vast portions of their collected data in order to steer the analysis to their desired conclusion. I absolutely stand by these statements because there truly isn’t a discussion in any portion of the authors’ papers explaining why they ever bothered to focus on the so called “M1” region of the brain. I have also gone on to explain in my review about how the mechanics of the study’s apparatus is the most significant factor in driving the participants’ hand movements and that the authors did not have a justifiable cause to eliminate data that did not match their notion of “accurate” movements as they unwittingly forced the “inaccurate” behavior onto the participants. A much more glaring mistake in their paper was that a set of hand movements in the “rotation run” physically did not correspond to any set of movements in the “baseline run” and yet the authors managed to find correlations between each set of movements across runs. Why is such a flawed paper lauded by the research community? I laid out a list of theoretical, methodological, analytical and mechanical flaws in the paper which was supposed to be the basis of a major portion of my PhD thesis as envisioned by my advisor who instead of acknowledging the list of concerns, retaliated in saying that I was not proceeding with the program’s objectives and was thus being let go.
I was being asked to ignore all the mathematics and physics I have been trained on as a systems engineer with a specialization in human factors just to promote a research program that is currently flush with money. The basic position of Prof Bapi Raju was that there is a popular way currently used by many people across the world who are able to conveniently ignore basic physics and go about doing “cognitive neuroscience” with fMRI machines so why was I being adamant about using laws of physics or a set of logical statements while going against those people who are funding my salary? He kept repeatedly insinuating that I did not appreciate the methods of empirical science and was being too philosophical and abstract. I was being blamed for being contemplative and analytical. Can it possibly be worthwhile to jump into an expensive and potentially hazardous experiment with an fMRI machine without adequate philosophical grounding nor a repository of fundamental principles? Oh wait, evaluating this question requires a philosophical starting point so it might as well be suppressed.
fMRI machines were not built by cognitive neuroscientists, nor by psychologists nor linguists. They were and are, built by engineers. The very concept of “functional” Magnetic Resonance Imaging came into existence at AT&T Bell Labs due to a direct application of something as abstract (tangible) as information theory which is actually a theorem pertaining to “concurrent changes within systems” and the theorem also explains what those changes can possibly “encode as information” . The limits of fMRI in providing a resolution to brain activity are also clearly laid out by Shannon’s information theory and Nyquist Theorem. Ignoring such limits possibly cannot yield valid research outcomes. But Prof Bapi Raju says that the applicability of physical law to study of cognition is a belief I harbor and because he does have a different belief system which need not incorporate my beliefs, I am wrong and must find employment elsewhere.
The ultimate consequence of this discussion was that I was removed from the program because I was not willing to compromise and be “flexible” enough to brush aside the flaws within a research program established via cronyism. The evidence of cronyism is in the fact that the most influential cognitive neuroscientists in this field of research who have exclusive access to high end fMRI machines collectively promote such unsound work that never stems from basic principles nor leads to basic principles i.e. produce a fundament understand of a phenomenon with an accurate account of the phenomenon. Of course, here we are trying to understand understanding which isn’t easy to bring into an empirical domain without various assumptions and approximations but such exorbitantly expensive research do not even lead to applications after those assumptions and approximations are slated (applications like designing rehabilitation regimes for those with a certain impairment). In such research it is observed that various parts of a brain exhibit metabolic activity while performing a muscular movement or while being subject to some stimulus. So what? More critically, these reported parts of the brain cannot be observed to have identical characteristics across individuals due to differences in brain structures, they cannot even be reported to have same characteristics within an individual over a period of time as the neuronal configurations are dynamic and evolve rapidly to showcase new characteristics. Furthermore, how can it be guaranteed that the combinations of low as well as high levels of BOLD signals are not relevant for a biomechanical movement and that only higher levels of BOLD signals analyzed via linearized models are most pertinent in explaining human cognition? How can such a dynamical system be explained by simplistic linear models, is a linear model even warranted?
Such questions are very important because all of the analysis done in the research I reviewed only looked at correlations between peak BOLD signals and experimental conditions matched within baseline run vs rotation run, averaged across trials as well as participants while eliminating sections of the rotation run data. This approach implicitly asserts that only higher levels of BOLD signals are relevant and that the very same brain structure across human beings yields the observed behavioral functions. Such an assumption thwarts the possibility of exploring the role of frequency encoding within asynchronous coupling among neurons or populations of neurons. It also eliminates the ability to study the possibility that different structures may yield the same function, a concept known as multiple realizability. Sabotaging the ability to explore a research avenue and subversively promoting a preconceived notion is plainly bad research methodology. This is a typical approach within current cognitive neuroscience research where the most influential scientists in the field have conveniently decided to gloss over as well as stay mum about the most critical issues related to the field while claiming to do important “empirical work”. Thus, evidencing cronyism through which a usurped title and position is consolidated while snubbing dissenting voices. At the end of such empirical work the standard conclusion is, “we need more money to do more research because we cannot, yet, provide accountable and sound statements about cognition”.
In this post I naturally sound rueful, overtly cynical and injured but I feel relieved that I have stood opposed to poor research practices that dismiss fundamental principles in the name of empiricism and waste research funding though I might not be able to prevent such wastage from continuing nor get anyone incharge at University of Hyderabad or Government of India or anywhere else to notice the fact that Indian and even European Tax Payer money is being squandered.
Figure taken from: Ben-Ari, Mordechai (2012). Mathematical Logic for Computer Science. Prentice Hall International Ltd: UK