Wednesday, February 24, 2016

What are the best interdisciplinary courses for undergraduate science majors?

Previously I posted about how science has changed [it is much more interdisciplinary and computational] and undergraduate science education really needs to catch up. I endorsed a great course on Physical models of living systems that Phil Nelson teaches and has just published a text for. [He kindly sent me a complimentary copy recently.].

In the recent UQ review of the B.Sc. there was some discussion of whether there should be more interdisciplinary courses offered and even whether one or two might be compulsory.
That got me thinking about what the best courses would be.

Energy and the Environment
David  MacKay at Cambridge has a book Sustainable energy: without the hot air

Advanced instrumentation and precision measurement
NMR, x-ray crystallography, laser spectroscopy, mass spectrometry, microscopy, ...
This should not just introduce these methods as "black boxes" but also describe the underlying science and let students have hands on experience.

Chemical biophysics or Biophysical chemistry
Either of Phil Nelson's two texts are idea models. On the other hand, they involve purely classical physics. One needs to also look at some quantum mechanics that relates to spectroscopy.

Computational biomolecular simulation and materials science
Again hands on experience is essential. But, a "black box" mentality must be avoided.

Materials science and engineering
There is a classic text by Callister, but a colleague tells me it is to much from the perspective of a metallurgist.

Here I think it needs to be continually emphasised that there is no point offering such courses if they are not rigorous and coherent. They need to be "owned" by just a few faculty. Courses will multiple guest lecturers from multiple departments quickly degenerate into a "dogs breakfast" and political nightmares. A single text is highly desirable.

What do you think?
What courses would you suggest?
What are suitable texts?


  1. I'm biased, but I think materials science is a good candidate. I like it for the following reasons:

    +Most science experiments involve materials
    +Materials are tangible and in your face
    +Because it covers the whole periodic table, there are more opportunities for global and comparative knowledge (like thinking in terms of organic materials vs metals vs semiconductors)
    +It applies to old technologies and new technologies alike
    +It doesn't overlap with high school courses (unlike intro physics/chemistry/biology)

  2. You note that science is becoming much more computational. I agree strongly with this observation.

    To make efficient use of software in the laboratory, it should be more than just functional; it should also be well-designed. Unfortunately, many graduate-level scientists are very poor practitioners of software design principles. From my own experience in a biophysics lab with an equal mix of PhD's and post-docs with backgrounds in physics, chemistry, and biology, I have observed that many young scientists and PhD candidates enter our lab knowing how to write code, but relatively few know how to design programs that are extensible, robust against bugs, and that facilitate reproducing analyses or simulations. Furthermore, there are few incentives at the PhD level in Europe to actually learn these skills during graduate school, possibly because of the limited time of the PhD (approximately 4 years); software design is perceived as something not to waste time on. The consequences of this thinking often lead to flawed analyses and difficulties in maintaining software tools in an environment with high personnel turnover.

    Luckily I think this state of affairs is changing in the greater scientific community. Data science has made very important contributions to both improving software tools and the perception that it is a necessary skill. The scientific suite built around Python is just fantastic and the community really encourages "doing things right," though I know other languages have made equal contributions to this area as well. In my own lab, everyone has started to realize how strongly we rely on code and software design, but we simply do not have the experience to capitalize on its use.

    To remedy this situation, I think at least one of two things must happen: serious efforts should be made to integrate this skill set into more facets of the undergraduate curriculum, or software design should be taught and enforced at the graduate level. Until the situation is remedied, I think that research at the graduate level will continue to suffer from inefficiencies and flaws associated with poor understanding of software design.

    1. Thanks for this very helpful and relevant comment.

  3. I think Energy and the Environment is a very good choice. Apart from rigorous thought about energy (which is sorely lacking in the public domain; the word "energy" is frequently misused because the physical concept of energy is not widely appreciated), it offers opportunities to prepare undergrads for work outside the academic world. Critical thinking about energy balances as well as associated societetal aspects from a scientific (or somewhat arrogantly from my side: physics) perspective is a very valuable thing.
    It teaches students to apply "ivory tower knowledge" to a practical thing - and one that is in the news daily and has/will have quite an influence on their lives the next 50 yrs.

    I also like advanced instrumentation; but it has to focus on the physics and technology behind the measurement. For students going the academic route it offers the opportunity to better understand how a measurement is done, and thus how to do it properly (and precisely) - this is fundamentally impossible in a black box explanation.
    For students not pursuing an academic career it again translates concepts in physics into knowledge of how this is applied in daily (technological) life. Even in (R&D in) industry, knowing the basics of how things are measured allows for more creative solutions to new problems. With only "black box" knowledge the default solution is to search for a vendor with another black box to measure a new thing. That won't give an economical edge.

    Finally, I like the materials science course. However, I dislike Callister - it indeed is written from a metallurgist point of view.
    In the end, I think this is a historical thing; materials science has a long history in metals, then came ceramics, then plastics. As a result the approaches (books) are fragmented in exactly these three fields.
    I think there is a good market for a book that is more generic "applied condensed matter physics" (which I think is exactly what materials science is). I do not know that such a book exists.
    [disclaimer: my perspective is colored by the fact that I did materials science as an undergrad and physics for my PhD]

  4. Forensic science.

    Bwahahahah just kidding.

    I think forensics wins the degrees produced vs jobs available ratio by a large margin.

    1. Thanks for the half-hearted suggestion. I agree Forensic science degrees are a silly marketing exercise. But I think one single semester course could be a nice one, particularly if it does not go the "black box" route on instrumentation.