Karl's End-SPUR presentation

[jensroes]

Karl will need to present the project at the end of his SPUR. We don't need to be stressed about this now and it is flexible when this can happen (ideally early in term 1). We can arrange that this is happening in the context of our "Language, Literacy, and Psycholinguistics" group. That's definitely a good context with friendly people that typically ask important questions. Alternatively we could look into presenting this in a uni-wide context: I'd think there is something like an event for UG presentations from different departments.

We'll probably want to present this on a larger conference as well such as the "International Conference on Interdisciplinary Advances in Statistical Learning" but this isn't happening before next year and my guess is that this is something that Linda, Mark or I should do.

Anyhoe, I thought it is useful to kick this off early as Karl probably wants to read a few related papers to get an idea of why he's getting people to look at dots. There is no need to start working on slides just yet.

One related read is

IAO, L.-S., ROESER, J., JUSTICE, L. and JONES, G., 2021. Concurrent visual learning of adjacent and nonadjacent dependencies in adults and children. Developmental Psychology, 57 (5), pp. 733-748. ISSN 0012-1649

Karl said he's already started reading it. This paper is the basis for why we're doing the current experiment.

Obviously Iao et al. (2021) used a different paradigm. To understand the bigger picture around the dot paradigm, I've sent Karl some relevant sections from a grant proposal that Gary Jones kindly allowed us to share (in parts). I didn't want to post this on github because it is confidential. I'd think that the proposal gives a good idea of why the dot paradigm is allowing us to answer new questions about learning of adjacent and nonadjacent dependencies and in particular when learning both adjacent and nonadjacent dependencies at the same time. Also, you will find up-to-date literature references in both Iao et al. (2021) and Gary's proposal.

An overview of the experiments we're (Karl is) currently running with our preliminary results can be found here: https://rpubs.com/jensroes/862437

@Lai-Sang is there anything you'd like to add which might be helpful for Karl to prepare the End-SPUR presentation?

[jensroes]

@karlnicholls I checked the SPUR ad and I could find any information about a SPUR conference. I might have missed something. If you find it, please post it here, so we have this on our radar.

[karlnicholls]

SPUR_project_adverts_2021_22..pdf I've highlighted the relevant section on page 2 in the PDF attached.

[jensroes]

After completion of projects, there is a SPUR conference where you will present your findings as a poster presentation. In addition, depending on the outcome of the work you may present the work at conferences and in some cases the research can lead to publication.

Okay thanks. Totally missed that. Sounds like this is expected, i.e. "where you will present", and I guess this can be fun. Even if we haven't got all the data yet, you can still use what we've got already to present this as a poster.

[jensroes]

@karlnicholls if you have access to the slides from yesterday, could you post them here? Would be good to have a reminder on when they want the poster to be presented.

[karlnicholls]

@jensroes I don't have access to the slides, but I recall them saying that the conference that the poster will be presented at would be the first week of November, but avoiding Bonfire night. I'm guessing the exact date isn't set yet.

[jensroes]

That's good enough for now. If you get an info about this, just leave a note here.

[jensroes]

I recall them saying that the conference that the poster will be presented at would be the first week of November

I just received a message of another SPUR student asking when they have to submit the posters to get them printed. Do you have any recollection about this. They said beginning September but that can't be true, right? If it is, we can just copy paste whatever we've already got from my rpubs page and put in some background. This isn't too difficult.

Just aside, there is now a tool to create posters in R: https://wytham.rbind.io/post/making-a-poster-in-r/

templates here: https://gist.github.com/Pakillo/4854e5d760351206084f6be8abe476b2

[karlnicholls]

@jensroes I checked the recording from the SPUR meeting and they said around the middle of October. The September thing was more of an "in an ideal world" sort of thing. They said as long as it's a few weeks before the event, then it's all good.

Those R tools could be useful, I'll definitely give them a go.

[jensroes]

Right okay, no idea why that would be an ideal world :) Does middle October work for you? I suppose you don't want to wait until you have finished data collection but use the results we've already available.

Also cool you want to use R markdown for the poster. I'm using markdown for all sorts of things, but I haven't tried the poster function yet, so I'm happy to learn. What might help is if I share with you / explain the R markdown code that generates the results overview you saw on rpubs. That would give you an idea how markdown works and you would already have the code for the plots and the numbers. You would then need to get the intro from the grant proposal snippet I shared with, the SPUR project description, and from Linda et al.'s paper. The intro needs to be very brief for a poster. Ideally you want to get across what research question we are tackling and why that is important. I think

[jensroes]

@karlnicholls how are you getting on with the poster? Do you need any input, comments on the poster?

[karlnicholls]

@jensroes I'm hoping to finish the poster over the weekend. @Lai-Sang sent me a few points for the introduction a while ago, but I'm not sure what more I need for it. I need the conclusion stuff and could do with guidance on whether/what to include in the analysis method section. I've highlighted what I've already condensed in green and what I'm yet to condense in red. I've also highlighted bits I'm thinking of cutting in yellow. Also, I need a title. Research Poster.pptx

[jensroes]

Could you send me pdf of this? I don't have MS word on my laptop so when I often a ppt file everything is all over the place :)

[Mark-Torrance]

PDF attached.

[jensroes]

PDF attached.

Sorry where?

[jensroes]

What I'd do judging from the ppt file:

• include the photograph you showed me
• for the plots in the design section increase the font size by changing base_size in the code I sent you to something larger than 14 probably in this bit

theme_set(theme_bw(base_size = 14) +
          theme(panel.grid = element_blank(),
                axis.title = element_blank(),
                axis.text = element_blank(),
                axis.ticks = element_blank()))

• remove the analysis methods section entirely. you don't need to explain this but you need to generally describe what the results show. people will not be familiar with bayesian models so just explain it in terms of main effects and interactions highlighted in green.
• results section: make plots and maybe table larger
• rearrange the participants section as a table so its easier to see the information
• make yourself first author for the poster
• the dependent variable needs to be explained in methods / design section (btw, call "design" section, "method"): time on target before target lit up
• title: Timecourse of concurrent learning of adjacent and nonadjacent dependencies
• add a findings section with these points: -- adjacent dependencies showed learning relatively early in the timecourse (E1, E2) -- no learning for nonadjacent dependencies (E1, E2) -- when presented simultaneously nonadjacent dependencies inhibited learning for adjacent dependencies (E3) -- explicitness of instructions did not seem to impact learning

Hope this helps for now.

[Mark-Torrance]

Sorry, here: Research.Poster.pdf

[Mark-Torrance]

@karlnicholls Do as much of the following as your feel like (some are quicker wins than others).

• Put yourself as first author. It's your poster.
• Remove all of the text highlighting.
• Explain adjacency with the simplest possible example: Control: random random random random Adjacent dependency: random A B random Non adjacent dependency: random A random B
• The language example is good - because it shows why we might be interested in this - but remove all of the jargon. Just give the example.
• You need to show the dot pattern, but just one is enough. I don't think you can tell adjacent and non-adjacent from your dot examples.
• Or, better, replace with a procedure diagram like this (or exactly this just credit Sofia Tsitsopoulou)
• The green and yellow highlighted bits (the bit that describes the manipulations) can be collapsed into...

All experiments - 2 blocks of 4 sequences of 3 elements, repeated 40 times per block. Experiment 1: One block of adjacent dependencies, one block of non-adjacent dependencies. Experiment 2: Same as Experiment 1 but participants explicitly told to predict next location. Experiment 3: Same as Experiment 2 but both blocks included both adjacent and non-adjacent dependencies.

• Make outcome measure really clear: Extent to which participant anticipates (predicts) next location in sequence, measured in fixations on next dot location in period before dot appears.

• Remove all text about statistical analysis. Just show plots.

• Summarise conclusions in two or three bullets. Keep this simple and jargon free.

• This will have won you loads of space. Keep total amount of text to 200 words (or similar) and increase font to a size that's easy to read from 6 feet away.

[Mark-Torrance]

Oh, and I wrote that without reading Jens' comments. I hope we agree. If not, I'm right.

[jensroes]

I hope we agree. If not, I'm right.

Unless I'm right. In this case though there is a lot of overlap and as for the things that don't overlap I agree with you :)

[karlnicholls]

@jensroes @Mark-Torrance Thanks for all of the advice. How is this looking?

Research Poster.pdf

I'm aware that the background section has single line spacing and the rest is 1.5 spacing, but if I make it 1.5 spacing there, I'm out of room. Also, does it matter that I have no references in the poster?

[jensroes]

For title and authors, can you try to use white font. I think white font is easier to see on purple background (similar to NTU logo).

For the research aim (last points in intro section), I'd put that our research aims to test learning of adjacent and nonadjacent dependencies individually and simultaneously (Iao et al. paper) and that we focused on the time course of learning either type of dependency.

I'd put references in very tiny font on the bottom but if there is no room it's fine without references, I think.

You can remove "Control A-B-C" cause the control (baseline) condition in the analysis is the transitions to the random dot (red in your sequences).

Remove the table label "Table 2.2".

It might actually help readability if you create some white space between individual sections. I'd image that all figures look huge in print (which is great), so you might be able to reduce the size a little. Also maybe you can simplify the text in the introduction a little.

Other than that, I think it looks great.

[karlnicholls]

@jensroes I've done most of what you've suggested, but I'm not entirely sure how to simplify the introduction bit further. I've had a go though. Also, I'm not sure where you meant with the Iao paper, I've put it where I thought you meant, but let me know if that wasn't correct. Research Poster.pdf

[jensroes]

Sorry, with "simplify" I mean turn sentences into phrases for bullet points and turn jargon into normal language but its fine as long as you can explain what statistical regularities and implicit learning means. (the trick is, instead of defining everything you're saying, just use words that people generally understand)

I'd remove the reference section. It just looks a little lost :) The citation can stay where it is.

Other than that I think it looks great. I hope we're too late with printing it. Do you know how to get it printed? I think the SPUR people sent out instructions.

[karlnicholls]

This was the email regarding posters. Maybe it's just a case of going to the print shop and saying it's for this event?

Shall we say this is the final version? Research Poster final.pdf

[jensroes]

Oops, just found the email. Apparently I've got to send it to the print shop :) I'll do that tomorrow morning, so @Mark-Torrance and @Lai-Sang any last suggestions now :)

[karlnicholls]

I don't suppose we know what the dress code for the event is?

[jensroes]

I don't suppose we know what the dress code for the event is?

I didn't expect that question.

My experience from psycholinguistics conferences is, the higher you're up in the food chain, the more you tend to wear shorts, an old t-shirt and flip-flops. Psychologists are very diverse from people that wear tie, shit and waistcoat every day and other people that are happy with t-shirts. Most staff will come to the event as part of their coffee break so they won't dress up unless they are always dressed up. So no need for tie and top-hat :) Simple but not fancy I guess.

[Mark-Torrance]

Normally mourning suits. But don't hire anything specially. A well-cut suit should be fine.

[jensroes]

Mark and I have very different ideas about this apparently.

[Mark-Torrance]

When I attend events of this sort I try and remember to change out of my sweaty running gear, and may even put on a clean t-shirt.

[jensroes]

So lets say no athletic clothing (e.g. running, hiking, riding, swimming gear)

[Mark-Torrance]

One of the reasons that I became an academic was so that I would never have to wear a suit to work. I am very suspicious of anyone who walks round universities in a jacket and / or a tie. Or, actually, even in a proper shirt. I feel they are hiding something.

[jensroes]

One of the reasons that I became an academic was so that I would never have to wear a suit to work. I am very suspicious of anyone who walks round universities in a jacket and / or a tie. Or, actually, even in a proper shirt. I feel they are hiding something.

I've said exactly that before, a lot, word for word. I agree especially with the last sentence. Apart from graduations, I've not come across a dress code at the university. For the latter, there is always a tie in my drawer.

[karlnicholls]

I didn't expect that question.

I like to ask the hard questions

So lets say no athletic clothing (e.g. running, hiking, riding, swimming gear)

So, leave the budgie smugglers at home? Got it.

One of the reasons that I became an academic was so that I would never have to wear a suit to work. I am very suspicious of anyone who walks round universities in a jacket and / or a tie. Or, actually, even in a proper shirt. I feel they are hiding something.

Despite how dapper he looks, Andrew Dunn must set alarm bells ringing for you then.

I was thinking more because this is meant to be a conference (albeit, more informal). I wouldn't want to arrive over/under dressed compared to others presenting.

[Mark-Torrance]

I was thinking more because this is meant to be a conference (albeit, more informal). I wouldn't want to arrive over/under dressed compared to others presenting.

In my experience some people dress up a bit for conferences. I tend to find a shirt with a collar. But in every conference I've been to you'd stand out if you wore a tie, for example. I would just wear what you'd normally wear. Personally, I also tend to avoid budgie smugglers between November and March.

[jensroes]

I've place the order (poster, not budgie smugglers). Lets see how quick they are :)