- Late Selection Theory:o Puts filter later on to try to help what the problem was in early selection theorieso Evidence supporting each theory but still missingo Start thinking about attention as a process, rather than a componento Shiffrin (computational mathematical psychology) and Schneider Has 2 papers, first is Shiffrin and Schneider, then the one that lays our empirical evidence and has Schneider as the first author Use terms for two types of processes used in the literature today Controlled and automatico Controlled processes: effortful, capacity limited, interference, serial operation, under conscious control Some resource that is of limited capacity, you can only do so manythings at one time, thought to be short- term memory Control processes make use of capacity limited resources, they willinterfere with each other, most efficient way to run them is to run them serially, which can be tested empiricallyo Automatic processes: effortless, not capacity limited, do not interfere, parallel operation, not conscious  Not using limited capacity, not constrained by the resources available, can run them parallel because they don’t interfere with each othero Shiffrin had main frame computers, no internal storage for these computers, storage was on tape that was on discs, 64K, o Visual search task- looking for target, appears on half of the trials, subject has to search for the target and press a key if present and a key if the target is absent, subjects do a lot of trials First a single letter appears on the screen, that’s what the target is for that trial, then the target is removed, and a x appears where the target was and the subject is told to focus on the x, the x disappearsand a circle of letter appears, x is there to help with reaction time and deal with extremely precise control, search depends on where you’re looking IV= number of items in the circle (search set size), reaction time increases at set size increases- Linear increasing- linear means time it takes is the same, increasing means its operating serially- Review letters in circle, find your target, make sure it is indeed your target- Target absent: cannot attend to more than one letter at a time, run it serially- Target present: depends where you start looking, if target appears equally often in all possible locations, than averagetime it finds to take target is half the time it takes to search through the entire displayo Linear increasing with slope of the reaction time function of target present is half the slope of the reaction time for target absent, 2:1 target absent:target present slope ratio---tells us we have a serial process DV= reaction timeo Empirical marker- 9/13/12- Parallel Process- absent and present happening at the same time, flat linear reaction time- Serial Process- 2:1 target absent, target present slope ratio- Assumptions you need to make for parallel and serial processes to mimic each other occurring is extreme, impossible to empirically test for this- Shiffrin and Schneider o Can something be changed from a controlled process to an automatic processo Target present: one letter is target, others are distracterso Target absent: all letters distractedo Done hundreds of times a day, multiple dayso In one of these groups for the entire training:o Consistent mapping- take group of letters, divide into 2 non overlapping sets, one is target set, one is distracter set, each target is consistently mapped and never changed, if you see letter as a target it will always be a target (same with distracters), target is drawn from same set- but is not always the same letter every time The letter always produces the same response, type of process going from controlled to automatic processing First day: everyone producing 2:1 ratio linear equation, at first everyone is serial processing After training: end up with flat parallel reaction time, indicating itbecomes automatic parallel processingo Varied mapping- all of the letters are in 1 set, both targets and distracters,any time you do the task the letter can play the role of a target or distracter After training: continued to produce 2:1 linear increasing functions9/18/12- Wouldn’t it be great if attention were automatized completely?o Couldn’t direct attention anywhere then- Bottleneck Theoristso Thinking about what mechanism it is- Shiffrin and Schneidero Each letter has own specific shapeo Attention starts out as a controlled processo After varied mapped training will stay controlled- Where do attentional processes operate? o At the shape or meaning? Sensory or semantic?o Schneider does an experimento Instead of using letters he uses words Order of letters is importanto Chooses 2 categories. Ex. Flowers and animalso Varied map people: any word can be target or distracter in any trialo Consistent map people: some targets, some distracters. Divides them by flowers and animals, all targets are flowers, all distracters animalso Category to which word belongs is being consistently mappedo Subjects go through training, get same results Consistently mapped people produce parallel reaction time Varied mapped people produce 2:1o Replaces flower and animal names with new oneso Sensory information is completely new- could not have developed automaticity to “daffodil” or meaning because they have never seen ito Consistent mapping: Should produce 2:1 ratio, because have no training DOES NOT HAPPEN When puts in new words: they continued to automatically respond to the new words, produce parallel reaction times- Early theorists looked at this as a component- Heavy reliance on metaphor- people tend to forget that it’s a metaphor, no computational model of the filter- No metaphor in Shiffrin and Schneider theoryo Capacity limited component- needs to be protected o You choose what gets let in, everything else gets blocked outo Everything in environment gets processed to some basic default level, very shallow, typically doesn’t rise to level of conscious awarenesso Attention is an active, go find stuff and bring it in