30 Matching Annotations
  1. Aug 2019
    1. ocietal impacts

      I'm picking two types of impact: technological and social. we can make new, better tools and that will improve people's lives.

    2. Speech is fundamental to human

      This is sorta dumb, but you need to situate your experiments in their broader societal relevance. In this case it's just so obvious that speech is important that saying so is sorta awkward.

    3. Computation in the brain is inherently dynamic, but models of dynamic computation are in their infancy

      And intellectual merits can be speculative -- this isn't a direct impact of this work, but is facilitated by it.

    4. Infant

      I think it's a good idea for your intellectual merit section to address impact for your field and other fields as I do in (3) and (4) here.

    5. I have successfully demonstrated that mice are capable of learning and generalizing consonant identityacross vowel contexts, speakers, and genders

      One reason I highly recommend doing the GRF in your second year is because you then have the strong advantage of pilot data. Even though your proposal is competing against a more established set of applications, I think (without data) having some proof of feasibility goes a long way towards demonstrating that you can actually get this shit done. You need to balance the perception that you've already 'done the project' with just letting them know you're capable of it by suggesting some next step that you need to do with the project.

    6. Two competing neurolinguistic theories

      There are, of course, more ways of conceptualizing this problem, but by setting up this dichotomy I give a framework for what I will be evaluating with my experiments. It's not dishonest to limit the space of discussion to answer a specific question. Notice how much space this takes -- only spend this much space on ideas that are the absolute core of your proposal. I repeatedly return to these ideas throughout the aims and final sections. Another thing this framing does is make it so my experiments will yield interpretable results no matter the outcome - there is no 'failure' to these experiments.

    7. Future Directions

      An additional, more general 'future directions' lets you touch briefly what could have been a third aim: you don't need the complexity of an additional aim, instead you just need to show you have options and other directions to take your project.

    8. Electrophysiologyand viral tracing are routine techniques in our lab.

      you have to justify your ability to actually do any methods you introduce - if your lab has their own 2p, put that in here, etc.

    9. Awake imaging during behavior gives two additionalexperimental tools

      In the last Aim I elaborated on some future directions, in this one I'm demonstrating that I know how to get the most out of my experiments, think flexibly about the use of different data that I get from it, etc.

    10. trials

      Make sure to leave AT LEAST 3/4 page for your intellectual merit/broader impacts section. giving the perception that you're shortchanging these sections or phoning them makes your proposal DOA.

    11. I will collaborate with Dr. Yashar Ahmadia

      This collaboration hasn't worked out (yet!), but they are on my committee and do advise me on the project. You don't need to hew to every detail in your proposal - grad school is long and science takes a winding path.

    12. standard meth

      "standard methods" - I don't really like this, and these techniques are very very very far from the only ways to model this data, but I make a compromise for the reviewer between depth of detail that demonstrates expertise (see my comments below the pdf, I don't think this is a good strategy) and sparseness of detail that may raise questions by leaning on collaborators in the next sentence.

    13. The discriminative model predictsinvariant patternsof temporalactivitywithin a region

      I can make this claim shorthand here because of the time i spent preparing the conceptual/analytical framework in the introduction.

    14. Because the two models make different predictions about the activity in a phoneme processing region, testing model predictions and identifying candidate regions will be a joint process.

      Here I need to describe what are, in reality, pretty technical analytical methods. There's a lot of detail that I could present to make my analysis more specific, but you just don't have space for that. Note my very qualitative description of the two competing theories, and how I am reusing & extending the framework I set up in the introduction.

    15. Dr. Kaori Idemaru, a phoneticianwho will aid in this process

      Mention collaborators whenever possible, especially if you're going a multidisciplinary project. Saying that you're working with experts in different fields really adds to the perception of feasibility - the reviewer can have faith that you're in good hands, and are good with reaching out for help.

    16. I will continue to validate the model by investigating the acoustic featuresmice use to discriminatephonemes, and compare these features to those used by humans

      I think it's smart to keep your aims short and specific, and include any further experiments as brief 'future directions' at the bottom of the aim. Of course, the whole experiment includes these two 'future directions' but by doing this you can give your reviewer a sense of how you're thinking about the problem without spending the space actually describing all your experiments.

    17. To our knowledge, this is the first time mice have been shown to be capable of learning phonetic categories, and,indeed,categoriesofnon-species-specific natural sounds in gen

      Be explicit (but humble) about what makes your experiment dope. Would yours be the first experiment to measure something? say that.

    18. nly a slightloss in accuracy

      If you are describing results, you do not need to actually present numbers. This isn't a paper for peer review, you just need to have your reviewers understand what it is that you're doing. Notice how I don't describe the experiment at all (what kind of task? how were they trained?)

    19. to “fill i

      Specifically colloquial "scare quotes" phrases can save a lot of space/confusion by replacing some jargon term with something understandable to a nonscientist/nonspecialist

    20. Aim1

      By giving the short description of the aims in the line above, I don't need to repeat them here.

    21. im 1:Establish mice as amodelfor speech perception. Aim 2:Identifycandidate cortical regions and processing mechanisms with widefieldcalcium imaging.

      This is a little cramped, so if you had an extra line I would split these out

    22. Thesemodels however, like all neurocomputational models of speech perception, are unconstrainedby neurobiological data

      Make a specific statement of empirical need: what is it exactly that we don't know about the question that you introduced at the top of the paragraph. the following sentence should tell the reviewer how you plan to answer that: ie. the one-sentence summary of your proposal.

    23. Coarticulation

      pick 1-2 jargon words (at the absolute max 3) if they are necessary for your proposal. explain them early, and use them consistently (ie. don't use both the expanded form of an abbreviation and the abbreviation).

    24. For example

      Examples are worth a thousand words. Taking the extra time here to give the reviewer something concrete to return to is very worth preventing the question from becoming muddled.

    25. We hear spoken words as a series of discrete sounds, or phonemes.Yet speech is continuous, andthe articulatory maneuvers required to produceone phonemepowerfully influence those needed to produce the next –they are “coarticulated.”

      You want to start by clearly presenting an open scientific question. Your reader should leave the first few sentences knowing exactly what it is that you propose to study. one common way is to present a dilemma/contradiction that motivates your project.

    26. 1. Kluender KR, et al.Vowel Inherent Spectral Change (2013), pp.117-151.

      You shouldn't use more than ~4-5 sources. I know it's the habit to be cited to the bone, but this isn't the venue for that. I recommend reducing the font size to 10, using numbered citations in the document, and then use this abbreviated citation style: "Last, first, et al. <bold>journal abbreviation</bold>. (year) volume(issue):pages

    27. Project Descriptio

      Don't waste space on whitespace, but keep your proposal readable. Double check on the formatting slack they they give you, but head your sections with bold, indented text but don't break the line afterwards. Give a few spaces or a colon and head into the paragraph

    28. 1)

      Don't be shy about numbering arguments in-paragraph, especially in the intellectual merit section. Listing several ideas in sequence makes your writing feel 'blippy' - moving too abruptly. By explicitly offsetting with numbers you implicitly prepare your reader for rapid topic switches. They also take up less space than "this, next, then, last, etc."

    29. I propose to developphonetic discrimination in mice as a model for speech perception. I will thendirectly testthe predictions made by competing neurolinguisticmodels

      End your introduction with a one-sentence description of your project. Use italics to make in-paragraph text easy for your reviewer to pick out.

    30. Introduction

      Space allocation - 1/3 page intro: you aren't writing an introduction that gets your reader knee-deep in the depth of your project. You want to give them just enough information to motivate your experiment without getting them bogged down in the details. This is easy to overdo - in general if you need more room than this than you aren't writing generally enough.