Guidelines for Coding the Object Examining Task

Adapted from a longer manual written by Jeni Swerdlow


We administer the Object Examining Task by having the infant sit on the parent's lap at a small table. Once infant and parent are seated, whoever is providing technical assistance must make sure that the camera angle is just right. The angle should be straight on, and close to the baby's face. Otherwise, if the camera is angled downwards, you can't tell what is being looked at precisely and can only see the top of their head when they look down at the stimuli. A downward angle makes coding nearly impossible. The camera lens should be aimed as directly towards the infant's face as possible. Fortunately, many infants pick up the object, thus raising it from the table, making their line of vision more straight into the camera lens.

Before actually beginning the task remind the parent not to label any of the toys, or direct their child's attention in any way. We also remind parents that their child's boredom with the objects we use is considered valuable information. These reminders make parents much less likely to bias their child. This is very important since biases will disqualify data.

By your side, you should have a tape recorder containing an audio tape that signals every twenty seconds with clicks. This clicking noise alerts the experimenter on-line that the trial is over, and it is time to switch exemplars. In addition, the clicks will afterwards allow a coder to mark the end of the trial (stop the stopwatch and record the time on the score sheet) during coding. Note also that several clicks arc on the very beginning of the tape giving the experimenter time to get ready to present the first object. You should also have a box nearby and out of the child's sight containing all six objects to be presented to the child. Strategically placing the items in the box helps remind the experimenter which object to present next. The box also helps to obstruct the infants' view, keeping them from peeking.

The test begins when the experimenter turns on the tape recorder and hears the first clicking sound. Upon hearing the first click, give the infant the first item. The infant keeps the item until the experimenter hears the second click, which is exactly when the second item is presented and first object is removed. You should always keep the next object to be presented in your hand on your lap, out of the child's view. That way, you are poised and ready to present the object as soon as you hear the click. In order to avoid a power struggle when removing the objects, wait until you hear the click, present the new item, then remove the old item. You may possibly need to gently pry the object out of the infant's hand. Usually, once an infant sees the next item on the table, they willingly relinquish the object. All of this happens in the blink of an eye and helps keep the infant on task. This procedure is followed throughout the experiment. Items 1 through 4 are presented once each for trials 1 through 4, then once each again for trials 5 through 8 (in the same order). Next present the two test items; the first one belongs to the same category as the familiarization objects and is trial 9. The other test item belongs to a contrasting category- this comprises trial 10.

If you are talking to the parent throughout the familiarization trials, you must continue to do so throughout the test trials. Similarly, if you are silent throughout the familiarization trials, you must also be silent during the test trials. The best option is not to talk, but sometimes this makes for a stiff and unnatural session. Talking to the parent has benefits however. The parent is distracted and is less apt to influence the infant. The infant, somewhat cognizant of being ignored, pays more attention to the objects. This can differ from subject to subject. We have also found that if the experimenter does not directly look at the infant it is better. Looking directly at the infant during the task tends to make the infant try to engage in some land of social interaction (smiling etc.). So, just look at the infant out of the corner of your eye, so that you do not become "irresistible". After all, infants can be adorable flirts. We also try to administer this task without any warm-up interactions. If you talk and coo with the infants before the task, they will get a little cranky if you all the sudden start ignoring them.

If you get involved in an object throwing game with the infant (one that is fascinated with dropping or tossing the object on the floor for you to pick up and return) there are several options you can try. Note that no single option is ideal and from time to time you will just have to forfeit data collection. One option is to retrieve the object and place it on the table slightly out of the infant's reach (but not completely). DO NOT LAUGH OR GIGGLE or you are sunk. Social reinforcement works very well, altogether too well. Do not look at the infant when you place the object back on the table. If the object is thrown again, gently place it again on the table this time further away and out of reach. Ignore struggles -- hopefully this will be a familiarization trial and loss of data on this trial can be accounted for statistically. Make sure you do not get yourself in the position of rewarding the infant's mischievousness (this is especially difficult if you are at the end of one trial and the beginning of another- you do not want it to appear that you are giving the infant a new object after another is thrown). If the parent reinforces such behavior try to divert the parent's attention.

Make sure that infants' tummies are snug against the table edge and the parent is not leaning back. You don't want the infants to put objects under the table or on their laps because you will not be able to reliably code their looking when the object is not in view on the film. Another trick is to get parents to put their arms outstretched on the table, one on either side of the infants, thus forming a mini-barrier so that items are not accidentally shoved off the table. We tell parents to do this if their infants begin pushing objects off the table.


In the past, in order to establish a definitive technique for scoring, we followed a detailed, lengthy procedure when coding what we have called the "look and toss" task or the "look and suck" task (for reasons that will become obvious to you as you test more and more infants). The coder would code the infant five separate times, each time coding for a single aspect of attention. First, the coder would determine the time the subject spent sucking the object. Then, the coder would do the same for time spent banging the object or waving it in the air. The next task was to code looking. Then, after looking was coded, the "quality" of looking behaviors was scored. Perhaps the coder noticed that some instances of looking were more intense than others, reflecting a focused concentration on the object. Finally, coders needed to go over the tapes once again and code "examining". By this time, coders were highly familiar with the subject's behaviors, and could anticipate one action from the next, and really take the time to further distinguish this specialized behavior. The coder gradually learned to look for specific actions while becoming more and more acquainted with a particular age and various behaviors. A last coding was done so that the coder could see how reliable the examining measure is in general. After reliability was obtained both within and between coders, coders could then score only examining behaviors.

Recently however, we modified the above training procedure because of several years of experience we've had using this measure. Now a lab assistant sits down with a new coder and both watch videotapes, while the lab assistant points out examples of examining. Using slow motion on the video player, it is easy to determine the start and stop of each examining period. This time together is a great forum for discussion, enabling the new coder to become familiar with the task and administration techniques. After this, the lab assistant gives the new coder the training packet, which includes blank score sheets for five subjects. These training tapes were specifically chosen for various degrees of coding difficulty. As you will discover, some infants are very easy to code and others are extremely difficult. The first two subjects are scored while the lab assistant sits nearby to offer advice and feedback. Once these two subjects are scored and at least 90% reliable with the original coders and the new coder, (see below for how reliability is calculated), the new coder independently scores the remaining three subjects. Again, these need to be at least 90% reliable before proceeding any further. We have found that most, but not all, undergraduate students are quite good at this.

When scoring, it is important to keep all of the following in mind:

* The coder will need a stopwatch, with a stop/reset function in order to determine
the examining time for each trial.

* After each trial, pause the tape and record the time on the score sheet. Trial end is noted by audio click that marks end of trial.

* Completely code all trials (items 1 through 10), REST, then code all trials again. Rest simply means get up, walk around, get a drink of water... Do not code the same trial two times in a row, because biases tend to set in. Code the subject on two independent occasions.

* If a subject mouths, sucks, bangs, or throws the object, it is not counted as looking time. Even though sucking can give the infant some information regarding the object's features, it requires cross-modal perception skills, which we can't evaluate, particularly in cases in which one item is visually examined and the next is sucked on.

* If an infant holds an object in the air as if handing it to parent/examiner, and may even appear to be looking at the item while doing this, this is not counted as looking time. The reason is that it is usually impossible for two coders to determine/agree if the infant is looking at the item or right past that item, to the person.

* Always code with the volume on so that you can determine the beginning and end of a trial, by listening for clicks from the videotape. Also, listen for any type of biases from parent or experimenter. "Look at that! Look here!", tapping the object, and labeling the object (which can inadvertently happen when the infant waves the object in the parent's face), instructing the infant to look at the item... Any and all of these should be clearly indicated on the coding sheet by the coder. Biases will disqualify data.

* Sometimes the size of the object can influence coding. If infants hold a larger object directly in front of their face, you need to determine if they are examining the object, or the object is simply in their plane of view. Again, it is important to consider "quality" of looking. Does the infant have a wide-eyed, glazed over look? Or is the infant engaged in inspection? It can be impossible to distinguish the two in cases where large objects are used. With large objects, the infants eyes appear wide open because a wide visual angle is required to look at the object as a whole, particularly if the object is held close to the eyes. Large objects can also get between the camera and the infants face, and make coding impossible. The ideal size of objects is too large to be swallowed or disappear into the infant's mouth, but small enough to be easily manipulated and scrutinized.


* Reliability with self-- calculated by the coder. After second coding is complete, coder verifies that the first coding and the second coding of each trial, are within one second of each another. Then, the coder calculates the mean of code 1 and code 2, which is what is used for between coder reliability.

* Between coder reliability -- calculated by the lab assistant. Between coder reliability for each trial is verified by determining that the difference between coder1(mean) and coder2(mean) is no greater than one second. Then, the assistant calculates % reliability for trials 8, 9, and 10.

Trials 9 and 10 are further examined to verify that both coders agree on which item was examined longer (was 9 examined longer than 10? or 10 longer than 9?). They must be in full agreement in terms of which was examined longer or they both must go back and recode the trials until they arrive at some agreement. It is noted on their coding sheets that there was a disagreement, and how it was resolved. Fortunately, this seldom happens.