EAM in Visual Perception Research

@47th ECVP 2025 Mainz

Carolina Maria Oletto

August 24, 2025

Why should we use EAM in visual perception?

Why should we use EAM in visual perception?

accuracy

Why should we use EAM in visual perception?

accuracy
RT

Why should we use EAM in visual perception?

accuracy
RT

Can we put them together?

Why should we use EAM in visual perception?

accuracy
RT

Can we put them together?

Speed-accuracy trade-off

Some examples

The effect of stimulus strength on the speed and accuracy of a perceptual decision

Which is the relation between RT and Accuracy measures?

Palmer et al., 2005

The effect of stimulus strength on the speed and accuracy of a perceptual decision

Which is the relation between RT and Accuracy measures?

Accuracy → Signal Detection Theory (SDT)
→ Psychometric function: d’ as a function of stimulus strength

Palmer et al., 2005

The effect of stimulus strength on the speed and accuracy of a perceptual decision

Which is the relation between RT and Accuracy measures?

Accuracy → Signal Detection Theory (SDT)
→ Psychometric function: d’ as a function of stimulus strength

RT → Chronometric function: mean RT as a function of stimulus strength

Palmer et al., 2005

The effect of stimulus strength on the speed and accuracy of a perceptual decision

Which is the relation between RT and Accuracy measures?

Accuracy → Signal Detection Theory (SDT)
→ Psychometric function: d’ as a function of stimulus strength

RT → Chronometric function: mean RT as a function of stimulus strength

RT and Accuracy depend on the difficulty of a perceptual judgment.

Palmer et al., 2005

Coupling of RT and Accuracy

DDM produces a fixed relationship between RT and accuracy for a given stimulus strength

SDT + separate RT modeling cannot capture this coupling
RT modeling alone ignores accuracy constraints
DDM integrates both, predicting how changes in stimulus strength shift RT and accuracy together
Single generative framework → fewer parameters, more precise predictions

Palmer et al., 2005

Drift rate reflects sensitivity to stimulus strength:

Increases with stronger stimuli

Robust across:
- Two response modalities (saccades and key pressing)
- Three different stimulus judgments (motion discrimination, contrast detection, contrast discrimination)

Palmer et al., 2005

Boundary separation reflects the speed-accuracy trade-off:

Larger boundaries produce slower but more accurate responses

Variations in instructions or conditions that prioritize speed versus accuracy are captured primarily by adjustments in boundary, without altering drift rate

Palmer et al., 2005

Dissociable mechanisms of speed-accuracy tradeoff during visual perceptual learning are revealed by a hierarchical drift-diffusion model

Drift-diffusion model to examine:

the speed-accuracy trad-eoff
perceptual learning effect

during learning of a coherent motion discrimination task across multiple training sessions.

Zhang et al., 2014

Boundary:

Larger under accuracy vs. speed emphasis
Decreases with training

Drift Rate:

Not significantly affected by speed-accuracy trade-off
Increases with training

Zhang et al., 2014

Enhancing change perception through object-based attention

In a change perception paradigm:

accuracy
RT

Is change perception facilitated by object-based attention?

Drift rate consistently higher for within vs. between conditions → faster evidence accumulation.

The Experiment

Stimuli

Manipulate difficulty to modulate:

Drift rate
Error rates (target: 5–35%)

Avoid:

Floor effects → guessing
Ceiling effects → no errors to fit

Boag et al. 2025

Stimuli

2 x 2 x 2 design:

2 gap sizes (small vs large)
2 conditions (congruent vs incongruent)
2 peripheral orientations (left vs right)

Response modality

EAMs assume the response begins after the decision ends.

Best modalities:

Manual keypresses
Saccades

Avoid imprecise, slow, or delayed responses.

Boag et al. 2025

Trial structure and event timing

EAM tasks follow a structured sequence of events:

Cue (optional)
Fixation
Stimulus onset
Response window
Intertrial interval

Each component affects the integrity of evidence accumulation.

Boag et al. 2025

Procedure

Task: targets (peripheral) lines orientation judgement left vs right.

Cue

Optional cue presented before stimulus onset.

Informs participants how to respond (e.g., emphasis on speed or accuracy).

Boag et al. 2025

Cue

May set cognitive control parameters:

Thresholds
Biases

Can direct gaze or attention to a spatial location.

Must occur before evidence accumulation begins.

Boag et al. 2025

Fixation interval

Ensures eyes and attention are centered.
Allows previous trial’s processes to return to baseline.
Reduces overlap across trials.

💡 Best practice: Use variable durations

Boag et al. 2025

Stimulus onset

Marks the start of evidence accumulation.
Assumes constant signal strength from onset to response.

Any variability or delay in onset weakens the assumption of continuous accumulation.

Boag et al. 2025

Response window

Starts with stimulus onset.

Ends with:

A response
Or a deadline

💡 Calibrate response window:

Long enough to allow natural responding

Short enough to avoid strategy shifts

Typical EAM use: mean RT < 1.5 s

Boag et al. 2025

Intertrial interval

Allow participant to reset

Prevent proactive interference

Avoid sequential effects

Boag et al. 2025

Collecting data

Participant ID
Condition
Stimulus presented
Response submitted
RT
Session/trial number
Event timings: cue, stimulus, response, feedback, intertrial interval

Boag et al. 2025

Hypotheses

Smaller gap enhances central interference (same-feature suppression)

Gap effect: small gap higher accuracy but slower RT (higher threshold)

Incongruent lines enhance central-peripheral segmentation

Congruency effect: incongruent conditions have higher accuracy and faster RT (higher drift rate)

Your turn now!