Pre-Determined Protocol ------------------------------------------------------------------------ **Subjects:** - MTurk sample - 90% approval rate to participate - USA only - pay $0.25 - N = minimum of 100/condition in final sample. Survey Monkey randomly assigns subjects to conditions, and that assignment can be uneven due to the random assignment and to differential completion rates across conditions. We will continue testing until we have a minimum of 100 subjects in each exercise condition (see stopping rules and exclusion processes below). - MTurk hits will be reposted until we reach the desired sample size **Tasks** *Note*: The complete survey used in this study is posted under Files and will be registered along with this protocol. It gives the details about what subjects read, see, and do. The survey was refined in the posted pilot studies to make sure that the attention checks worked and to make sure the data were interpretable. *Exercise Intervention*: Subjects will read about one of two fitness plans, with each subject randomly assigned by SurveyMonkey to that condition. One condition describes an aerobic fitness plan consisting mostly of walking 3 times/week. The other involves stretching/toning/weights 3 times/week with different exercises introduced. Both interventions last for 6 months. *Cognitive tasks*: Each subject will then read a cognitive task and will watch a 1-minute videos showing the tasks. They then will judge whether or not the exercise intervention would improve task performance, and if so, by how much. The order of the cognitive tasks is randomized by SurveyMonkey for each subject. *Demographics*: Subjects will provide basic demographic information about themselves (e.g., age, sex, frequency of exercise, etc). They also will note their familiarity with research on the effects of exercise on cognition. *Attention checks*: After the exercise description and after each cognitive task, subjects are asked to describe the intervention/task they just learned about. These attention checks make sure that subjects understood the description and will also serve as a means to exclude subjects who did not. **Stopping Rules and Exclusion Principles** *Stopping rule*: Data collection will cease after we have obtained usable (non-excluded) data from 100 or more subjects in each exercise intervention condition. This number of subjects is the same as we have used in other studies of differential expectation effects that found reliable effects. If this sample size proves underpowered (e.g., we get an uncertain estimate of the size of the intervention effect), we will conduct a new study with greater power and will not add additional subjects to this one. We will determine whether or not to exclude subjects during data collection. It is possible that we might collect more than 100 subjects in each condition if additional data comes in while we're determining which subjects to exclude. If so, we will use all usable data. Critically, we will not look at responses to the critical dependent variables until data collection ceases. *Exclusion due to incomplete responses*: Subjects who do not complete the full survey will be excluded from analyses. *Exclusion due to lack of comprehension*: Usable data means data for which the subject demonstrated clear understanding of the intervention and of each task description in the attention check questions. To avoid biased exclusion, DS will create a stripped version of the data file that includes only the attention check questions and not the primary dependent measures (i.e., the ones asking subjects whether they would expect improvements). CS and WB will independently examine those responses and will flag subjects who gave inaccurate or incomplete descriptions of the intervention or tasks. CS and WB will know which exercise intervention description the subject read and will know which task description they read, but they will not have any information about whether that subject thought the intervention would improve performance. They then will return their separate spreadsheets to DS without communicating about any individual subjects. If either CS or WB judges any of a subject's answers to reflect incomplete or inaccurate comprehension, data from that subject will be excluded from analyses. In this manner, we can ensure that all subjects included in the analysis read the descriptions of the intervention and the tasks and understood them before making their judgments about improvements. *Exclusion due to familiarity with exercise research*: The survey asks participants if they are familiar with research on the effects of exercise on cognition. If they answer yes, they are asked to describe what they know. Based on pilot data, approximately half of the participants had heard that exercise can improve cognition and gave some description of the types of exercise that can improve cognition. This question comes at the end of the survey, so participants are likely to be led by the earlier questions and the intervention described to them. Moreover, many of the pilot subjects gave vague responses that did not reflect understanding of the specific effects of exercise on cognition or that made a blanket statement that did not differentiate types of exercise (e.g., "exercise can keep the mind sharp"). Given that the central hypothesis is about differential expectations for the two types of exercise interventions, we will exclude participants who differentiate between the effects of cardio/aerobic and strength/toning on cognition in their answers to these questions (e.g., "cardio exercise improves cognition") or who describe an effect associated with just one of the types (e.g., "some exercises increase blood flow to the brain which improves cognition"). DS will create a stripped version of the data file that includes only the answers to these questions about exercise research familiarity (along with the attention check measures above). CS and WB will code each subject's responses, and if either indicates that the subject differentiated between the effects of cardio/aerobic and strength/toning on cognition, we will exclude that subject from the primary analyses. Note that we may include these subjects in subsequent exploratory analyses, but they will not contribute to the data constituting the minimum sample size. **Analyses of critical hyptheses** The primary analysis will compare the expected improvements across the two exercise interventions. The literature shows that cardio/aerobic interventions (i.e., walking) lead to greater cognitive improvements than do stretching/toning/lifting interventions. The question here is whether people expect cardio/aerobic interventions to have a greater impact on cognitive tasks (when they aren't familiar with that particular finding). If they do, then the differential improvements following an actual intervention could be due to differential placebo effects. For each cognitive task, we will compare the proportion of subjects in each exercise condition who expect improvements from the intervention. That analysis will consist of separate Chi-square tests for each cognitive task comparing the cardio/aerobic condition to the stretching/toning condition. For the subset of subjects expecting an improvement, we will conduct a t-test to compare the extent of improvement expected. We will separately analyze data from sedentary subjects, those who report not exercising. We will conduct the same three Chi-Square analyses and t-tests for that subset of the sample. This analysis will test the possibility that expectations differ for those who don't exercise given that the actual interventions used sedentary subjects. This analysis might be underpowered depending on the proportion of respondents who exercise. **Planned exploratory analyses** We will examine whether those who exercise expect bigger improvements than those who do not exercise. We will also explore whether the type of exercise people do predicts their expectations for improvement. That is, do those who do cardio/aerobicexpect bigger improvements from cardio/aerobic than do those who do stretching/toning.