Miami University

ROBIN THOMAS

INTRODUCTION TO SEQUENCED WRITING ASSIGNMENTS

The following project was designed to incorporate several methods and modeling tools that are the focus of my course in mathematical psychology (Psy 373).  I have developed this course as a “tools” course that allows students to acquire techniques of mathematical modeling (including working with analytical representations as well as computer implementations and analysis software) in the context of the behavioral sciences.  Previously, when I taught this course, I had the students perform the technical tasks (collecting data, analyzing, etc.) but did not have them do significant writing; or if they did write, it was simply a final report for which I did not require drafts.  After my experience in the CWE workshop on writing in content courses, I redesigned this project emphasizing writing tasks along the way to help them understand the concepts as well as the mechanics of modeling.

The biggest changes from previous versions of this assignment are the inclusion of smaller assignments of a developmental nature (Assignments 1,2, and 3) prior to the final paper (Assignment 4) and the inclusion of a specific grading rubric which makes clear how their papers will be evaluated.  I hope that these changes will highlight the use of writing as a process of understanding to achieve my course objectives rather than simply a means of assessment of student performance.

Robin D. Thomas

Department of Psychology

SEQUENCED WRITING ASSIGNMENTS FOR STUDENTS

Psy373 Mathematical Psychology

Similarity Project - A sequence of assignments

Project objectives:  The purpose of this project is to go through the entire process of conducting an experiment, applying a model to the obtained data from it, and reporting the results.  After this project you will be able to:

1. interpret patterns in data from the perspective of a formal model
2. apply the technical tools of analysis in applying models to real data, and
3. write a technical communication to other scientists of your findings.

Context:  What effect does learning to categorize have on the perceived similarity of objects?  Specifically, does how we perceive an object’s attributes and its relations to other objects change once we learn that some objects belong to one category while others belong to another?  We will attempt to answer this in four assignments.  The reading for this project is Schyns, et al (1998).

Overall Project Outline

Assn. 1:  Construct and research stimulus materials (Small paper plus stimuli)

Assn. 2:  Obtain baseline similarity ratings and their MDS configuration (Analysis and small paper)

Assn. 3:  Perform categorization task and second similarity ratings with their MDS configuration (Analysis and small paper)

Assn. 4:  Integrate findings from assignments 2 and 3 into research report (Full research paper)

 Assignment 1: Development of stimulus materials

Think of a set of objects that fall naturally into two categories, yet are, overall, fairly similar to each other.  For example, mushrooms can be either poisonous or edible, yet it is difficult for novice mushroom hunters to tell the difference.     Do a little research on their physical characteristics including identifying what information experts may be using to distinguish them.  Select a set of at least 20 (half in one category, half in the other) whose pictures can be mounted on 4x6 index cards. Write a brief description of your objects and the nature of the categories (and their differences).  This should be no more than one or two paragraphs long and include any citations to the source for identifying category membership.

Assignment 2: Assess the baseline similarity and representation

Find a volunteer and ask them to assess the dissimilarity between pairs of your objects (1 = most similar, 7 = least similar).  Use the schedule of presentation order given in class for different numbers of stimulus sets.  Do not tell your volunteer that the stimuli fall into two categories yet.  Here you just want to evaluate how they are perceiving the stimuli without any specific experience with them (Note:  make sure your participant is not coincidentally an expert in dealing with your stimuli!).  Analyze the obtained dissimilarity ratings using MDS in SPSS.  Specifically, perform an ordinary Euclidean MDS selecting the appropriate number of dimensions using the elbow criterion.  Report the results of this analysis including all the plots of interest (e.g., stress plot, stimulus space plot).  Can you identify what dimensions your observer might be using to compare the stimuli?  Are the objects dispersed according to their category or are they randomly mixed together in the stimulus space?  Your report should be approximately 1 – 2 pages for this including plots.

Assignment 3: Categorization and post-similarity assessments

Using the same volunteer, have them learn to categorize your objects using the concept identification paradigm discussed in class.  Once they can successfully go through the set 3 times without error, redo the dissimilarity ratings task from Step 2.  Perform an ordinary Euclidean MDS on the post-categorization dissimilarity ratings.  Again,  where do the objects fall: according to their category, or are they randomly mixed together in the stimulus space?  Your report should be approximately 1 – 2 pages for this.

Assignment 4: Global analysis and overall report

Now fit an individual differences MDS using both the pre and post categorization dissimilarity ratings as your two “subjects”.  That is, put the (lower half) matrices, one on top of the other, in the data window of SPSS and run an MDS choosing Individual Differences as the scaling model.  Again, choose the appropriate number of dimensions using the elbow criterion.

Writing assignment:  You are to write a report arguing either for or against the hypothesis that learning to classify alters the psychological representation, and hence, the perceived similarities of objects in the categories.  You will also need to interpret any changes in perceived similarity in light of the demands of categorization and expertise.  Include any plots, stress values, etc., you think help your case.  Here are some questions to help guide your thinking about how the participant’s representation may change (use them to help you think about the problem, do not answer them one after the other specifically).

1. Does the dimensionality in Assignment 3 change from Assignment 2 (more dimensions or fewer?).  If so, how?  Are there more dimensions post-categorization, or fewer?

1. Does the configuration (i.e., locations of points in the space) from Assignment 2 look similar to that in Assignment3? Does it look similar to the group configuration in Assignment 4?  

1. Looking at the individual differences solutions, are the weights in the post-configuration different from the weights in the pre configuration?  If so, how?  Does the change make sense given what is required for classification?
2. Do you see evidence in your data for the development of features that experts might use discussed in Assignment 1?

In order to enhance the quality of the final report, I will require a draft one week prior to the final report’s due date.  This draft will count for 10 of the 50 points for the final report.

Suggested structure of the Final Report:

This follows the APA style in general form.  It is to be typed, double-spaced with 1inch margins.   If you know that style from previous classes, use it.  I will require drafts of sections (see below and timetable) to encourage revision before the final product is due.  For overall format, you should include the following components:

Introduction:  Here, introduce the reader to the issue of category learning potentially changing how we perceive object properties.  Use Schyns, et al (1999) paper as a reference for this.  Orient the reader to the present study by indicating what tasks you plan to perform to address the issue.  For this include:

1. A description of your stimuli (generally) and how experts identify category membership

1. A statement of your hypothesis (what you expect category learning will do to the perception of attributes)

1. A discussion of the distance-based model of similarity as a means of discovering these effects

Methods:  Describe, in detail, the stimulus set, the tasks, and the procedure of the entire experiment. 

Results and Discussion:  Report the statistical analyses (include tables and plots as necessary.  Here, discuss the effects (or non-effects) of category learning that you observed using your thoughts of the above questions as a guide.  Specifically address whether your data provide evidence in support of your hypotheses stated in the introduction.

References:  If you used references, provide a citation list at the end.  A correct citation for a journal article looks like this:

Schyns, P. G., Goldstone, R. L.; Thibaut, J. (1998) The development of features in object concepts. Behavioral & Brain Sciences, 21, 1-54.

For other types of sources see the APA Publication Manual or the online Citation Machine found on the Library website (Miamilink).

Timetable for completion and credit portions:

Evaluation

Attributes of importance for short components: (Assignments 1, 2, and 3)

Clearly stated purpose – what is the point of the writing?

Well developed context – why should the reader be interested?

Accuracy – are the statements made correct (e.g., literature attributions, results of analyses)?

Style and mechanics – is the writing organized and free of grammatical, syntax, and spelling errors?

Attributes of importance for final paper:

Introduction

• How well are the motivation and background context developed?
• How clearly stated and cleverly embedded within this background context is the hypothesis?
• How adequately is the current experiment forecast?

Methods

• How accurately are the stimuli and tasks articulated – well enough for replication?

Results and Discussion

• How clearly are the results of the analyses presented?  Is the MDS model of similarity sufficiently articulated that the reader can understand your results?
• Are your analyses accurate (i.e., correctly done and appropriately graphed)?
• How well do you connect the findings to the hypothesis and context laid out in the introduction?

References

• Are they accurately cited?

Organization and Mechanics

• How well do the ideas flow within the paper (good transitions, logical sequencing, etc.)?
• Are you using an appropriate writing style for an audience of fellow scientists?
• How good are the mechanics (grammar, syntax, spelling)?

Grading Rubric for Final Paper