Poelstra, Sanne (2021) What Can We Learn from Absent Cues? Master's Thesis / Essay, Human-Machine Communication.
|
Text
mHMC_2021_PoelstraS.pdf Download (6MB) | Preview |
|
![[img]](https://fse.studenttheses.ub.rug.nl/style/images/fileicons/text.png) |
Text
toestemming.pdf Restricted to Registered users only Download (122kB) |
Abstract
Error Driven Learning (EDL) is a theory of learning that states that we learn by using stimuli (cues) to predict certain outcomes. EDL is often represented by a simple neural network with cues as input, outcomes as output and weights as predictions. Theories of learning on how to update those weights exist and they differ in their implementation. Rescorla and Wagner (1972, RW) state that absent cues (cues that have been seen before, but are not seen now) should not lead to an update of the weights, while Van Hamme and Wasserman (1994, VHW) propose that absent cues should update weights. In this thesis, we modelled the experiment from Van Hamme and Wasserman(1994) with both the RW and VHW algorithms. Although these algorithms make different predictions in certain circumstances, the original experiment does not seem to tease these two predictions apart. For that reason, we conducted three experiments. We found support in the direction of the Rescorla-Wagner model, thus indicating that learning might not occur in the absence of cues. We will discuss the results in terms of task effects on implicit learning versus explicit inference and how this aspect could be addressed in future research.
| Item Type: | Thesis (Master's Thesis / Essay) |
|---|---|
| Supervisor name: | Rij-Tange, J.C. van |
| Degree programme: | Human-Machine Communication |
| Thesis type: | Master's Thesis / Essay |
| Language: | English |
| Date Deposited: | 20 Jul 2021 13:09 |
| Last Modified: | 20 Jul 2021 13:09 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/25353 |
Actions (login required)
 |
View Item |