Skip to content

Visual Search Experiment - Pausch et al. (1997) VR Study Replication

This example contains a full SightLab implementation replicating the landmark study:

Quantifying Immersion in Virtual Reality
Randy Pausch, Dennis Proffitt, George Williams (1997)


Download Latest Version

Study Overview

This experiment replicates the classic visual search paradigm used to measure how immersive interfaces (VR Walking Navigation vs. Desktop Controller) influence search efficiency.

Participants are tasked with finding a specific letter in a room filled with distractors.

Experimental Variables

  • Navigation Condition:
    • Walking (Physical head-tracked movement)
    • Desktop with Controller
  • Trial Types:
    • Red Y (Target Present, feature pop-out)
    • Black Y (Target Present, conjunction search)
    • Black Y (Target Absent) (50/50 chance of the target letter not existing)

Dependent Variables

  • Reaction Time (RT)
  • Fixations/Saccades
  • View count sequence structure
  • Walk Path Patterns

Response Options

  • 1 or 2 keys
  • Likert Scale after 10 seconds forcing a choice

How to Run

Requirements

  • Vizard 7+
  • SightLab Toolkit installed

To Run the Experiment

  • Manually input whether the group is 'VR' or 'Desktop' and whether it is 'uncamouflaged (Target letter is red)'
  • Modify experiment.py with the following values:
    • PRACTICE_TRIALS = 0 → Set to 0 to skip practice
    • REAL_TRIALS = 10 → Number of Trials
    • CONDITION_GROUP = 'Desktop' → 'Desktop' or 'VR'
    • UNCAMOUFLAGED = False → True or False for camouflaged or uncamouflaged
    • NUM_DISTRACTORS = 85 → Or any number you want
    • TARGET_ITEM = 'Y' → Target letter or item to find
    • ITEM_SET = list('AKMNVWXYZ') → List of items including distractors
    • RESPONSE = keypressResponse() #Waits for the 1 or 2 key. Option 2- delayedRatingResponse()
    • INSTRUCTIONS = Can put custom instructions here
  • Make sure VR hardware is set up and ready if using the VR condition
  • Run Main_Pausch_Visual_Search in Vizard or by double-clicking
  • Follow on-screen instructions

Data Storage

  • Experiment Metrics:
    Stored in the /data/ folder using the default SightLab data structure, which includes core gaze and navigation metrics as well as reaction time and response accuracy

Data Analysis

A VisualSearch_Comparison_Analysis.py script is included to analyze performance across conditions (for all sessions in the data folder by default).
It compares:

  • Navigation Condition (Walking vs Controller)
  • Reaction Time & Accuracy

You can extend the analysis to include view count, dwell time, fixation patterns, and more.

What the Experiment Does

The participant will:
1. Read instructions
2. Complete randomized trials
- Each trial displays the target type.
3. Keypress Option: Press '1' if they found the target, '2' if it was absent
Likert Option: Wait 10 seconds and forced to make a choice
4. Letter positions and response times are logged

The design replicates the findings from:

Pausch, R., Proffitt, D., & Williams, G. (1997).
Quantifying Immersion in Virtual Reality.
Proceedings of the 24th Annual Conference on Computer Graphics and Interactive Techniques.

SightLab Data Collection

All experiment data is automatically saved in the /data/ folder following SightLab's standard format, which includes:

Core Eye Tracking Metrics

  • Timestamp
  • Trial Label
  • Gaze Intersection (Combined, Left & Right)
  • Eye Euler Angles (Combined, Left & Right)
  • Head Position (6DOF)
  • Fixation / Saccade State
  • Saccade Amplitude & Velocity
  • Pupil Diameter (device dependent)
  • Eye Openness (device dependent)
  • Dwell Time
  • View Count
  • Fixation Count
  • Time to First Fixation
  • Reaction Time
  • Condition & Custom Trial Data

Object-Specific Gaze Data

  • Dwell Time per Object
  • Fixations per Object
  • Total & Average Gaze Duration per Object
  • Fixation Sequence Analysis
  • AOI (Area of Interest) Metrics

Physiological Correlates (Optional)

When paired with BIOPAC systems:
- Electrodermal Activity (EDA)
- Heart Rate (HR)
- EEG, fNIRS, and others

Extended SightLab Capabilities

This experiment can easily be extended to include:
- Rating / Likert Scales
- Multi-User Collaborative Search
- Real-time Biofeedback Display
- Adaptive Trial Difficulty
- Custom Event Markers
- Gaze Contingent Displays
- Replay Heatmaps, Scan Paths, Walk Paths

All of these can be configured using SightLab's GUI or script API.

Video Sample

Full Code

"""
This experiment tests visual search performance under two conditions (VR vs Desktop)
and target-present vs target-absent trials. Optional to have target camouflaged or uncamoflouged
Structure:
1) Instructions
2) Optional: Practice Trials
3) Running Trials (with randomized blocking, stimulus presentation & judgment)
4) Conclusion
"""

import experiment
from experiment import showInstructions, runPracticeTrials, runTrials, endExperiment
from experiment import *
import sightlab_utils.sightlab as sl
from sightlab_utils.settings import *

sightlab = sl.SightLab(gui=False)
sightlab.setStartText(' ')
env = vizfx.addChild('Resources/whiteroom.osgb')
sightlab.setEnvironment(env)
sightlab.setTrialCount(experiment.TOTAL_TRIALS)

def sightLabExperiment():
    yield viztask.waitEvent(EXPERIMENT_START)
    yield showInstructions(sightlab)
    yield runPracticeTrials(sightlab)
    yield runTrials(sightlab)
    yield endExperiment(sightlab)

viztask.schedule(sightlab.runExperiment)
viztask.schedule(sightLabExperiment)
viz.callback(viz.getEventID('ResetPosition'), sightlab.resetViewPoint)

experiment.py

import viztask
import random
import viz
from Modules.letter_module import createItems, clearLetters, saveLetterPositions
from Modules.response_module import keypressResponse, delayedRatingResponse

PRACTICE_TRIALS = 0  # Set to 0 to skip practice
REAL_TRIALS = 10
CONDITION_GROUP = 'Desktop' #Set to VR or Desktop
UNCAMOUFLAGED = True
NUM_DISTRACTORS = 85  # or whatever number you want
TARGET_ITEM = 'Y'  
ITEM_SET = list('AKMNVWXYZ') # Or filenames of 3D models
RESPONSE = keypressResponse() #Waits for the 1 or 2 key
#RESPONSE = delayedRatingResponse() #Waits 10 seconds and shows a rating

TOTAL_TRIALS = PRACTICE_TRIALS + REAL_TRIALS

def chooseTrialType():
    if random.random() < 0.5:
        if UNCAMOUFLAGED:
            return {'target': TARGET_ITEM, 'color': viz.RED, 'present': True, 'uncamouflaged': True}
        else:
            return {'target': TARGET_ITEM, 'color': viz.BLACK, 'present': True, 'uncamouflaged': False}
    else:
        return {'target': TARGET_ITEM, 'color': viz.BLACK, 'present': False, 'uncamouflaged': False}

def showInstructions(sightlab):
    pass

def runPracticeTrials(sightlab):
    if PRACTICE_TRIALS > 0:
        for i in range(PRACTICE_TRIALS):
            nav_condition = CONDITION_GROUP
            trial_type = chooseTrialType()
            yield presentStimuli(
                sightlab, trial_type, nav_condition, i, 
                record_data=False, response_handler=RESPONSE
            )
        # Practice complete message
        yield sightlab.showInstructions("Practice complete.\n\nPress Trigger to begin the real trials.")
        yield viztask.waitEvent('triggerPress')

def runTrials(sightlab):
    for i in range(PRACTICE_TRIALS, TOTAL_TRIALS): 
        nav_condition = CONDITION_GROUP
        trial_type = chooseTrialType()
        yield presentStimuli(sightlab, trial_type, nav_condition, i, response_handler=RESPONSE)

def presentStimuli(sightlab, trial_type, nav_condition, trial_index, record_data=True, response_handler=None):
    trial_label = 'Practice' if not record_data else nav_condition
    trial_type_text = "Practice Trial\n\n" if not record_data else ""
    start_text = f"{trial_type_text}Find the Y\n\n Press 1 if present, 2 if not\n\nPress Trigger to Continue "

    # ---------------------------
    # Fixation or Instruction
    # ---------------------------
    yield sightlab.startTrial(
        startTrialText=start_text,
        trialLabel=trial_label,
        trackingDataLogging=record_data,
        experimentSummaryLogging=record_data,
        replayDataLogging=record_data,
        timeLineLogging=record_data
    )

    # ---------------------------
    # Stimulus Presentation
    # ---------------------------
    letters = createItems(trial_type, sightlab, ITEM_SET, NUM_DISTRACTORS)

    if record_data:
        saveLetterPositions(letters, trial_index + 1)

    start_time = sightlab.hud.getCurrentTime()

    # ---------------------------
    # Judgment (Discrete DVs)
    # ---------------------------
    response_key = yield response_handler(sightlab)

    rt = round(sightlab.hud.getCurrentTime() - start_time, 4)
    expected_response = '1' if trial_type['present'] else '2'
    correct = (response_key == expected_response)

    print(f"{'PRACTICE' if not record_data else 'TRIAL'} {trial_index + 1}: "
          f"TargetPresent={trial_type['present']}, "
          f"Pressed={response_key}, Expected={expected_response}, Correct={correct}")

    if record_data:
        sightlab.setExperimentSummaryData('NavigationCondition', nav_condition)
        sightlab.setExperimentSummaryData('TargetPresent', trial_type['present'])
        sightlab.setExperimentSummaryData('ReactionTime', rt)
        sightlab.setExperimentSummaryData('ResponseCorrect', correct)

    clearLetters(letters)
    yield sightlab.endTrial()

def endExperiment(sightlab):
    conclusion_text = "Thank you for participating!"
    yield sightlab.endTrial(endExperimentText=conclusion_text)
Was this page helpful?