1
00:00:02.920 --> 00:00:11.280
Good morning to everyone. I'm Davide Bacchin, and I will present our work. Gaze-based metrics of cognitive load in a conjunctive visual memory task.

2
00:00:11.780 --> 00:00:20.720
The objective of our paper is measuring cognitive load during this novel task the CVSTM, which exploits conjunctive feature to overload the visual short-term memory.

3
00:00:21.700 --> 00:00:25.750
The literature represents many ways to measure cognitive workload.

4
00:00:25.920 --> 00:00:35.710
Among them, objective measures based on psychophysiology offer reduced intrusiveness, real time assessment, and the analysis of interpersonal variability.

5
00:00:36.620 --> 00:00:51.760
In our work the aim was to validate a new experimental paradigm called CVSTM, and its ability to induce different levels of cognitive low through the exploitation of indexes based on eye-tracking method, which provides reliable and easy to access data.

6
00:00:52.270 --> 00:01:03.290
Among the possible indexes, we selected two novelty ones.  LHIPA that is based on the Fourier transformation of the signals derived from pupil circulation, and microsaccade magnitude.

7
00:01:03.410 --> 00:01:24.230
We believe that this task could better approximate usual tasks in human computer interaction studies requiring visual, short-term memory, such as interface evaluation.  To compare the ability of CVSTM to evoke different levels of cognitive load, we compare it to the well known n-back task with n used to evoke different levels of cognitive load.

8
00:01:24.290 --> 00:01:33.140
In the slide, the reader can visualize this procedure, and the stimuli provided. The 1-back task was considered the easy level, while the 2-back, the hard one.

9
00:01:34.480 --> 00:01:50.640
In the CVSTM, the task required remembering the location and color of a series of squares, which were 4 in the easy condition, and 8 in the hard one.  The participants has to indicate if the prompt presented the same 2 features of one of them.

10
00:01:51.970 --> 00:01:55.180
The procedures is schematically shown on this slide.

11
00:01:56.690 --> 00:02:15.510
Accuracy measures suggest that a multi-valent feature of the easy CVSTM task indicated as significantly harder than the monovalent nature of of the easy n-back, agreeing with subjective impressions derived from NASA TLX questionnaire of significantly increased difficulty of the former versus the latter.

12
00:02:16.900 --> 00:02:24.390
This data shows that LHIPA discriminates between baseline and task condition, but not between difficulty levels.

13
00:02:24.690 --> 00:02:42.190
Results show LHIPA indicating n-back as the harder task and the reverse for microsadde magnitude. One possible explanation is that LHIPA is more sensitive to tasks involving short-term memory, while microsaccades are more sensitive to a visual short-term memory.

14
00:02:42.800 --> 00:02:54.870
It is possible to conclude that CVSTM can be a viable tool to evoke mental effort in scenarios approximating real world Human Computer Interaction in conjunction with the use of eye tracking methods.

15
00:02:55.050 --> 00:02:56.600
Thank you for your attention.