Background
The purpose of the Brown and Hulme (1995) model was to show that item length (or word length) effects could be obtained in a model that does not include any rehearsal.
Items are made up of segments. For simplicity, it is assumed that the initial strenght of each segment is 0.95. Each segment takes 0.1 s. A word with 4 segments would thus last 0.4 s; another way of saying this is that in 1 second, 2.5 4-segment words could be spoken.
All items are presented at a rate of 1 s per item, so for a list of 4-segment items, there is 0.6 s interval between each word. For a list of 7-segment items, there is a 0.3 s interval between each word.
Instructions
Click once on the "Run the Program" button. You will see a screen that looks like the following (the version number and date that your copy of the program was compiled may be different):
Brown and Hulme 0.5 08 Jan 2002 07:47 EST Enter the starting number of items [1-10, 3]
This is how the model prompts for intput. The first number is the suggested value, and the second part is the range of allowable values. So, the default value is to begin with a 3 item list, but you can specify any list length between 1 and 10. If you enter a value that is not legal or is out of range, the question will be repeated.
This implementation will compute performance for 5 list lengths at a time beginning with the length you specify.
You will then be prompted for more information:
Enter the number of segments [1-10, 4] Enter input decay [0.0-1.0, 0.0001] Enter output decay [0.0-1.0, 0.001] Enter initial trace strength [0.0-1.0, 0.95] Enter forgetting probability [0.0-1.0, 0.1] Enter length dependent recall increment [0.0-1.0, 0.036] Enter length independent recall increment [0.0-1.0, 0.164] Enter L (minimum segment) [1-5, 4] Display full results [1=Yes, 0=No]
When the simulation finishes, a new window will appear. If you use the default values and do not display full results, the window will contain something like the following:
Brown and Hulme 0.5 08 Jan 2002 07:47 EST Caution: Simulation is accurate enough for demonstration purposes but is *** NOT *** sufficiently accurate for scientific research. Parameter Settings: numItems = 3 segments = 4 inputDecay = 0.00010 outputDecay = 0.00100 initialStrength = 0.95 forget = 0.1 ldri = 0.036 liri = 0.164 ell = 4 List Length = 3 Decay times for each segment after presentation: Segment 1 Segment 2 Segment 3 Segment 4 Item 1 29.00000 28.00000 27.00000 26.00000 Item 2 19.00000 18.00000 17.00000 16.00000 Item 3 9.00000 8.00000 7.00000 6.00000 Recall probability for each segment after presentation: Segment 1 Segment 2 Segment 3 Segment 4 Item 1 0.94725 0.94734 0.94744 0.94753 Item 2 0.94820 0.94829 0.94839 0.94848 Item 3 0.94915 0.94924 0.94934 0.94943 Recall probability for each segment after output decay: Segment 1 Segment 2 Segment 3 Segment 4 Item 1 0.94725 0.94734 0.94744 0.94753 Item 2 0.94347 0.94356 0.94365 0.94375 Item 3 0.93970 0.93979 0.93988 0.93998 Item probability (product of segment strengths): Item 1 Item 2 Item 3 0.80560 0.79280 0.78021 Item probability (after forgetting and increments): Item 1 Item 2 Item 3 0.88904 0.87752 0.86619 Recall probability of entire list: List Length 3 0.67575 Decay times for each segment after recall: Segment 1 Segment 2 Segment 3 Segment 4 Item 1 0.00000 0.00000 0.00000 0.00000 Item 2 5.00000 5.00000 5.00000 5.00000 Item 3 10.00000 10.00000 10.00000 10.00000
Suggested Simulations
What follows are suggestions for simulations to run and how to run them. Unless otherwise stated, the values used for each parameter are the default values.
Simulation 1
Length and Lexicality (Demonstration 1, page 602). Compute span for short, medium, and long familiar words, for short, medium, and long unfamiliar words, and for short, medium, and long unfamiliar nonwords. Short items have a speech rate of 2.5 items per sec (0.4 s duration), medium items have a speech rate of 2.0 items per sec (0.5 s duration), and long items have a speech rate of 1.43 items per sec (0.7 s duration). L will be 4 for all simulations.
Familiar words: LDRI = 0.036 LIRI = 0.164 Unfamiliar words: LDRI = 0.03 LIRI = 0.14 Unfamiliar non-words: LDRI = 0.00 LIRI = 0.12
You should get something like the following:
Note: Original indicates the results from the 1995 paper; Web indicates the results from the current simulation version.
