Who says science can't be cool.

-->

Welcome to Enkhbold Nyamsuren's page.

My Google Scholar


Email: e.nyamsuren[AT]gmail[DOT]com

Last updated on 2018.04.16



Pre-Attentive and Attentive Vision Module for ACT-R

Module version. This is a beta release, and not yet a full version of the module. Future releases may include changes in both syntax and semantics.

 

Contributors to module's development.

Last source code update: 2014.01.08

Version: 0.98e

 

Lisp code for the module with VSTM

 

 

Last source code update: 2014.09.17

Version: 0.98g

Description: This is an experimental version. For a more stable release, use above version 0.98e

 

Lisp code

 

 

Supported functionalities.This extension for ACT-R's default vision module includes following new functionalities:

  • Defining visual object's properties along five feature dimensions: color, shading, shape, size and orientation.
  • Differential acuity of object's feature perception.
  • Iconic memory.
  • Bottom-up and top-down mechanisms for guiding visual attention.
  • Visual decision threshold.
  • Variable landing point for a gaze.
  • Variable saccade duration based on saccade's amplitude.
  • Visual regions.
  • Visual acuity support for a text.
  • Location information for visual objects is now persistent once visual object enters iconic memory.
  • Low-capacity high-resolution Visual Short-Term Memory, or VSTM for short (call (sgp :enable-vstm t) to use this functionality).
  • Ability to create abstract/imaginary visual object inside VSTM (call (sgp :enable-vstm t) to use this functionality).

Installation instructions. To install the module, simply copy the source code file into "modules" folder of ACT-R. Tested with ACT-R 6.0.

 

ACT-R source code. This is an official distribution of an ACT-R package with which PAAV is guaranteed to work. Release version is r1227, June 4 2012.

 

Download ACT-R package

 

 

User manual. Coming up NOT so soon! (sorry for that)

 

Human Reasoning Module (HRM) for ACT-R

Module version. This is a prototype release, and not yet a full version of the module. Future releases will include changes in both syntax and semantics.

 

Last source code update: 2014.01.15

 

Download module's lisp code

 

 

Supported functionalities.

  • Triple-based knowledge organization inside declarative memory.
  • Function-free horn rules for expressing conditional declarative knowledge.
  • Reasoning pipeline.
  • Bottom-up spatial reasoning with Visual Short-Term Memory.
  • Top-down reasoning via rule-based inference.
  • Production rule-based implementation of backward-chaining approach.

Module requirements. PAAV module v0.98e (with VSTM enabled) is required for full HRM functionality. Download PAAV module code above.

 

Installation instructions. To install the module, simply copy the source code file into "modules" folder of ACT-R. Tested with ACT-R 6.0.

 

ACT-R source code. This is an official distribution of an ACT-R package with which HRM is guaranteed to work. Release version is r1227, June 4 2012.

 

Download ACT-R package

 

 

ACT-R model of Math Garden's set players.

Module requirements. PAAV module 0.99g is required.

 

An updated version for the Cognitive Science journal. The package below contains human and model data, trials' descriptions, ACT-R model, and R code for statistical analysis. Update date: 2015.06.18.

 

Download package

 

 

An older version of the model for the ICCM2015 conference. Update date: 2015.04.06.

 

Download model code

 

 

Trial descriptions, user and model data for ICCM2015 conference.

 

Download data

 

 

ACT-R model for a Casual Deduction task.

Module requirements. HRM module is required.

 

Last source code and data update: 2014.01.14

 

Download model's act-r code

 

 

Experiment trials. Description of trials used by the model is embedded inside model's code.

 

Results from model runs. Download results from 50 model runs.

 

Download model results

 

 

ACT-R model for a Spatial Relations task.

Module requirements. VSTM and HRM are required.

 

Last source code and data update: 2014.01.14

 

Download model's act-r code

 

 

Experiment trials. Description of trials used by the model is embedded inside model's code.

 

Results from model runs. Download results from 50 model runs.

 

Download model results

 

 

ACT-R model for a Blicket task.

Module requirements. HRM module is required.

 

Last source code and data update: 2014.01.14

 

Download model's act-r code

 

 

Experiment trials. Description of trials used by the model is embedded inside model's code.

 

Results from model runs.

 

Download model results

 

 

ACT-R model for Feature and Conjuction search tasks.

Module requirements. PAAV module, either v0.98d or v0.98c, is required.

 

Model. This single model can do both tasks.

 

Last source code and data update: 2012.06.26

 

Download model's act-r code

 

 

Task trials. Download set of trials we used to test the model.

 

Download trials for feature search task

 

Download trials for conjunction search task

 

 

Results from model runs. Download results from model runs on visual search trials provided above. The results include RTs, model's fixation records and feedback records.

 

Download results from feature search task

 

Download results from conjunction search task

 

ACT-R model for Comparative Visual Search task.

Module requirements. PAAV module, either v0.98d or v0.98c, is required.

 

Model.

 

Last source code and data update: 2012.06.26

 

Download model's act-r code

 

 

Task trials. Download set of trials we used to test the model.

 

Download trials for cvs task

 

 

Results from model runs. Download results from model runs on visual search trials provided above. The results include RTs, model's fixation records and model's feedback records.

 

Download results from cvs task

 

ACT-R model for a game of SET.

Module requirements. PAAV module v0.98c is recommended.

 

Model. Given 12 cards, one of them being highlighted, the model should find other two cards that form a set with the highlighted card.

 

Last source code and data update: 2012.06.26

 

Download model's act-r code

 

 

Task trials. There are 30 different trials. Trial descriptions are embedded inside the model code.

 

Results from model runs. Download results from 50 model runs. The results include RTs and model's fixation records.

 

Download model results from set task

 

 

Results from an experiment with human participants. Download results from 14 human subjects who did the same 30 trials as the model. The results include RTs and subjects' fixation records. Those are the raw ASCII versions of EDF files produces by SR Research EyeLink 1000 tracker.

 

Download results from SET experiment

 

 

Areas of interest. Download a file describing areas of interest used to annotate model and subjects' fixations.

 

Download AOI file

 

ACT-R model for a game of Word SET.

Module requirements. PAAV module v0.98c is recommended.

 

Word Set Model. Given 12 cards, one of them being highlighted, the model should find other two cards that form a set with the highlighted card.

 

Last source code and data update: 2012.06.26

 

Download model's act-r code

 

 

Picture Set Model. Given 12 cards, one of them being highlighted, the model should find other two cards that form a set with the highlighted card.

 

Last source code and data update: 2012.06.26

 

Download model's act-r code

 

 

Task trials. 32 trials of picture set and 32 trials of word set.

 

Download trial descriptions

 

 

Results from model runs. Download simulation results of both picture and word set models. The results include RTs and model's fixation records.

 

Download model results from set task

 

 

Results from an experiment with human participants. Download results from 18 human subjects who did the same 64 trials as the two models. The results include RTs and subjects' fixation records encoded as original EDF files produces by SR Research EyeLink 1000 tracker.

 

Download results from SET experiment

 

 

Areas of interest. Download a file describing areas of interest used to annotate model and subjects' fixations.

 

Download AOI file

 

ACT-R model for a MAV experiment.

Module requirements. PAAV module v0.98c is recommended.

 

Model. Given 12 set cards, the model should identify the most abundant value, a value that occurs in largest number of cards.

 

Last source code and data update: 2013.02.01

 

Download model's act-r code

 

 

Experiment trials. There are 336 different trials that were used with both experiment with human participants and ACT-R model simulation.

 

Download trials.

 

 

Results from model runs. Download results from 50 model runs. The results include model's fixation records and responses.

 

Download model results from MAV experiment

 

 

Results from an experiment with human participants. Download results from 14 human subjects who did the same 336 trials as the model. The results include subjects' responses and fixation records. Fixation records include the raw ASCII versions of EDF files produces by SR Research EyeLink 1000 tracker and annotated fixation sequences.

 

Download results from MAV experiment

 

 

Areas of interest. Download a file describing areas of interest used to annotate model and subjects' fixations.

 

Download AOI file