Preface

Welcome to this course on Sensation and Perception (aka, S&P)!

This is one of my favorite classes to teach. Over the years, however, I have become increasingly frustrated by the rising cost of college textbooks. This is especially true for Sensation and Perception, where textbooks can easily cost well over $100.

Unfortunately, I have never found an open-source S&P textbook that fits the way I teach this course. To address this, I originally assembled individual book chapters from a variety of sources and made them available to students through our Moodle site (affectionately known as Lyceum here at Bates). While this was certainly better than asking students to purchase an expensive textbook, it was far from ideal. The chapters varied in writing style, organization, and level of detail, and they did not always match what I emphasized in class. Quite honestly, I'm not sure how many students actually read them or found them especially useful.

So, what to do?

When generative AI tools such as OpenAI's ChatGPT became available, it occurred to me that they might provide a solution. I found that asking ChatGPT to "write a Sensation and Perception textbook" produces terrible results. However, during the COVID-19 pandemic I recorded lecture videos for all of my classes, and those videos included closed-caption transcripts.

For the first edition of this book, I provided ChatGPT with the transcript of each lecture along with a prompt similar to this: Use the text of the following lecture to create a chapter entitled [chapter name] that could appear in a college-level textbook. Provide details and incorporate all of the points discussed in this text.

The resulting chapters served only as a starting point. I then carefully edited the text, corrected inaccuracies, reorganized material, improved explanations, and added figures that I either created myself or obtained under Creative Commons licenses.

During this process I also learned an important lesson about AI: it can be remarkably useful, but it can also be spectacularly wrong. At one point ChatGPT somehow generated an entire section discussing greenhouse gases while summarizing a lecture on auditory perception. I still have absolutely no idea how that happened.

The second edition of this textbook represented another large step. After using the textbook for a semester, I carefully reread every chapter and performed extensive revisions. Many sections were rewritten, expanded, reorganized, or clarified until they better reflected my lectures and the way I explain these topics in class.

New in Version 3

This latest edition goes even further. While the first edition began as AI-generated summaries of my lectures, the textbook has evolved beyond that original draft. Nearly every chapter has been substantially revised, expanded, reorganized, and clarified. I have also added references throughout the book. In class, I usually explain what researchers discovered without stopping to mention who discovered it, so those citations never appeared in the lecture transcripts that served as the starting point for the first two editions. This edition adds those references so you can more easily trace ideas back to the original research. However, one of the most valuable resources for improving the book has been my students.

Throughout the semester I maintain an online discussion forum where students can post questions while studying for quizzes and exams. These questions provide wonderful insight into which concepts students find confusing. Rather than simply answering the questions online and moving on, I decided to use them as a guide for revising the textbook.

As a result, this edition includes substantial additions and clarifications on numerous topics, including: psychophysical methods, Stevens' power law, pupil size and attraction, the blind spot, rods and cones, horizontal cells, Mach bands, the Hermann grid illusion, simultaneous lightness contrast, White's illusion, cells in primary visual cortex, 2-Deoxyglucose (2-DG), neural fatigue, cortical magnification, double dissociations, the dorsal and ventral visual pathways, saccades, change blindness, inattentional blindness, Feature Integration Theory, Gestalt grouping principles, object substitution masking, top-down influences on object recognition, the Dress illusion, color processing within the lateral geniculate nucleus, opponent-process theory, the Ames room illusion, stereoscopic depth perception using red-blue glasses, motion-sensitive neurons in area MT/V5, corollary discharge theory, neural mechanisms of motion perception, ways in which motion can disrupt conscious awareness, sound frequency, amplitude, and loudness, the Laurel/Yanny illusion, Shepard scales, the tritone paradox, auditory pathways, Fourier analysis, synesthesia, the historical significance of the thermal grill illusion and the work of J. Henry Alston, phantom limbs, thermoreceptors, pain perception, mechanoreceptors, nociception, 6-PROP and different types of tasters, the basic taste qualities, substances that alter taste perception, olfactory pathways, the basic categories of smell, and pheromones.

Many of the figures have also been updated. This edition also includes 37 new figures, several of which were generated using AI and then further refined in Photoshop. While AI is remarkably good at generating images, it also has an unfortunate tendency to produce incorrect labels, anatomically impossible bodies (extra fingers and arms seem to be a favorite), and other amusing errors. Every figure has therefore been reviewed and corrected before being included.

Use of AI

In my opinion AI is an extraordinarily useful tool, but it is just that, a tool. AI can accelerate writing, generate illustrations, and serve as a useful starting point. At the same time, it requires knowledgeable human oversight. Facts must be verified. Explanations need refinement. Figures need correction.

This naturally raises another question: if your professor can use AI to help create a textbook, can students use AI on assignments?

The answer depends on the guidelines you have received from your professor, so read your syllabus and ask questions if anything is unclear.

However, a useful guide is to think of the acronym P.L.A.C.E.B.O.

P = Prompts. The quality of AI output depends heavily on the quality of your prompts. The more specific your prompt, the better the results are likely to be.

L = Learning objectives. AI should never be used in ways that undermine the learning objectives of the course.

A = Accuracy. You are responsible for verifying everything you submit. AI can produce convincing but incorrect information, and spotting errors is often easiest for someone who already understands the topic.

C = Cumbersome. Sometimes AI actually makes writing harder. It is often quicker to write simple ideas yourself.

E = Ethics. If you are instructed not to use AI on a particular assignment, then using it would be considered unethical and would constitute academic dishonesty.

B = Bias. AI systems may reflect biases present in their training data. Always consider whether bias may have influenced the information that was generated.

O = Origin. The work you submit should fundamentally originate from you. AI should assist your thinking but should not replace it. If your prompts supply the critical ideas and content, AI is helping communicate your work rather than creating it for you.

If you use AI in ways that violate the policies described in your course syllabus, point deductions or other academic consequences may result (e.g., and these could be severe including failure of the course). If you are ever uncertain whether AI is appropriate, please talk with your instructor.

Finally, although this textbook closely follows the material presented in class, it is certainly not a substitute for attending class. The lectures include many demonstrations, discussions, examples, and additional details that simply cannot be captured in a textbook.

I hope you find this book useful. It has been an evolving project and I fully expect it to continue evolving in future editions. If you notice an error or think a concept could be explained more clearly, please let me know. Many of the improvements in this edition came directly from student questions, so your feedback will very likely help improve the next edition.

I hope you enjoy the course!

Todd A. Kahan