Serial Order Basics

The Serial Order Task

animal presented with 5 stimuli

must respond to the stimuli in correct order

less than a 1% chance the animal gets the correct order

animals task several months to learn this task and you must start with one stimulus and then add the rest over time

Forward Error

selects a stimulus too early, it should appear later in the pattern

Backward Error

goes back to a stimulus that was already pressed

Pairwise Test

with 5 stimuli there are 10 possible pairs

after the animal reaches criterion with the serial order task, they are presented with this task

animal now receives two stimuli instead of 5 - will it respond to them in the correct order based on training?

outcome: both monkeys and pigeons will do well in any pair that has an A or E in it - monkeys can perform the internal pairs but pigeons cannot

First Item Latency Effect

measured on a pairwise test

when the two stimuli appear, how long does it take the animal to respond to the first item?

outcome: monkeys show the first item latency effect, they are quicker to respond to a pair with A and linearly slower to B, C, and D - this is because they are thinking through the sequence every time so it takes longer to get to D than it does to get to A - pigeons do not show this effect meaning they are not going through the list and likely just guessing on the middle pairs

Missing Item Effect

how long does it take the animal to respond to the second stimulus after already responding to the first

outcome: quicker to respond to pairs that are right next to each other like AB or BC because there is no missing item - slower to respond to AE - monkeys show increased latency as the amount of missing items increases because they are going through the whole sequence - pigeons do not display this effect, confused by the middle pairs

Symbolic Distance Effect

also based on latency to the first time

but this time we are looking at the latency of each different pair compared to the other pairs with the same first stimulus (AB vs. AC vs. AD vs. AE)

is the latency to A in AB different to the latency to A in AE

outcome: A is such a powerful cue that it won’t have a symbolic distance effect - monkeys are fastest on BE, slower on BC and CD - BE is wider apart and this it is easier to differentiate between the two stimuli than compared to BC or BD - no evidence that pigeons show this effect

Serial Order: Humans vs. Monkeys

method: humans and monkeys taught five item serial order test and given pairwise test

pairwise test outcome: both do well, humans do a bit better

first item latency effect: both display a first item latency effect

missing item effect (pairwise test): humans don’t show this because they are thinking logically, simply press the remaining item, monkeys show this effect

missing item effect (triplet test): generates missing item effect in humans on the second item

symbolic distance effect: both show this effect

Pigeons and the BD Pair: Method 1

method: train on ABCD or ABCDE - instead of pairwise test just give them pair BC/BD - BC+ means reward for responding B then C, BC- means reward for responding C then B - if pigeons are clueless about the positions of B and C then both groups should learn at the same speed

outcome: BC+ and BD+ learn serial order task much quicker than the BC- and BD- - this means pigeons do have a sense of what comes first and what comes second

Pigeons and the BD Pair: Method 2

method: train pigeons on ABCD - embed probes of BC pairs into the regular ABCD trials - this doesn’t shock the pigeons like the pairwise test

outcome: pigeons do very well, no problem responding to the BC probes when they’re embedded into the training session

Chunking Study

method: trained birds on serial order task with lists that either could be chunked or were less likely to be chunked - five groups with different stimuli (all color, 3 color 2 form, alternating, 4 color 1 form, 2 color 1 form 2 color)

overall speed of learning the task: lists that were chunkable were learned must faster than unchunkable lists

speed of learning each phase of the serial order task: naturally takes longer to learn the serial order task the more stimuli you add - the chunked groups have a different pattern based on the chunks

performance on pairwise test: unchunked groups bomb the middle pairs - the chunked groups do well on middle pairs that happen to be at the start/end of a chunk because they are functionally end items

Abstract Numerical Reasoning Study

method: pigeons trained on a variety of three-item lists using numbers 1, 2, 3 - saw different stimuli but always played with a 1,2,3 order - tested on novel stimuli but some had familiar numbers (1-3) and others had unfamiliar numbers (4-9) - must respond in numerical order - if a pigeon gets 4 and 8, does it know to respond to 4 then 8?

pair type vs. percent correct: monkeys and pigeons perform well on FF, FN, and worse on NN but still good

numeric distance vs. percent correct: as the distance betwen the numbers increases, you do better (1 vs 7 is easier than 6 vs 7), shown in both monkeys and pigeons

response latency vs. numeric distance: as the numeric distance increases, response latency decreases - the wider apart two numbers are, the quicker you can respond - shown in both monkeys and pigeons

numeric ratio vs. percent correct: as numeric ratio increases, percent correct decreases - discriminating between 1 and 2 (1/2) is easier than discriminating between 7 and 8 (7/8) - lowest numeric ratio would be 1 and 9 - lower numeric ratio means easier to distinguish between two numbers