Attention has been studied heavily and as per a popular model of attention by Posner et al, we have 3 systems for attention: alerting, orienting and an executive control network.
Now let me propose a radical fourth network for the same, but before I do that I want to clear some misconceptions about attention. Way back in 2009 I had blogged and elaborated on a couple of posts that attention allocation and action selection utilized the same mechanisms and were conceptually similar; at that time I was not much aware of the Posner et al model of attention. Today I want to go further and claim that attention processes are involved in action initiation and selection.
But lets start from first principles.
We can always be in either an alert state on the lookout for stimuli or in a more sleepy/drowsy state where we will probably ignore stimulus, the extreme being when we are sleeping and ignoring all stimuli. This system is also know as arousal system and is fairly unequivocally associated with Norepinephrine (NE) system. Tonic NE levels (tonic meaning the baseline/spontaneous firing rate of NE neurons in LC (locus coerelus) ) as per one theory drive the activity of this alerting network in the brain. The higher the tonic NE and more alert you are; lower the NE and you get drowsy /sleepy. Too much NE/alertness and it actually becomes distractability where you cannot remain focused on task at hand but get distracted/ alerted by each and any irrelevant stimuli. So the relationship of alertness/ tonic NE with task performance follows inverted U of Yerkes Dodsan law.
So to detect a particular stimulus (of relevance) the first step is to be on the lookout for stimuli in general. And alerting system does this beautifully – it provides a knob to tune whether you want to ignore most stimuli or to attend to most stimuli in your current state. By playing around with tonic NE levels that can be easily accomplished.
Once you are in varied state of readiness to detect stimuli in general (on various levels of alertness), you may be primed to attend to a particular spatial location or a particular modality for a particular stimulus. On ANT (attention network test) this is accomplished by providing a spatial cue that indicates where the target will appear and primes the human/animal to turn its gaze either covertly or overtly to that location. In real world phenomenon, some CS will predict that an UCS is going to follow and teh animal/ human will react by directing attention to the location/ modality of that UCS prediction. For eg., when you hear thunder, you will be on lookout for a lightening visual to follow and will be using orienting attention to become primed for the same. Typically, orienting attention is based on prediction: there are hundreds of locations one can attend to, but due to predicting cues one orients to a particular location in space, in anticipation of a stimulus of relevance.
So the second step to detect a stimulus of relevance is to orient one’s attention keeping in mind the preceding cue(s). Its not clear whether this is accomplished by tonic NE levels or phasic NE levels, but what is known is that this is a faster response as indicated by pupil diameters (of eyes) which are somehow associated with NE levels and dilate faster in orienting than in alerting.
And once you are both alert and oriented then the stimulus of relevance appears (or doesn’t).
Once the stimulus appears, it rarely appears alone. Only in contrived lab settings does the stimulus appear alone; and even there in some conditions it is flanked by flankers (distractors) which can be either congruent or in-congruent. So we need to suppress irrelevant stimuli and attend selectively to relevant stimuli. Now one can debate what is relevant- for the purpose of this section it is the stimulus on whose lookout we re as per the previous step of orientation. Because it is surrounded by irrelevant background information/ stimuli executive networks kicks in and suppresses the flankers/ irrelevant information.
So the third step in detecting a stimuli is a ‘hot’ process where we latch onto the relevant stimuli and suppress irrelevant stimuli. And this is accomplished beautifully by phasic NE firings which ensure that once a decision is being made, it is widely amplified and winner takes all happens.
However the story is not complete yet.
The final stage in detecting a stimuli is doing deep processing of selected stimuli (a ‘cold’ process) and determining what action to take / response to met out. This is akin to final selection of stimulus as worthy of action and further processing.
So to summarize:
- Readiness to detect stimuli (alerting)
- Priming for a particular stimulus (orienting)
- Stimulus appears
- Suppressing irrelevant stimuli (executive control – conflict )
- Selecting final stimuli for deep processing (executive control- decision)
So the above is the way in which action happens in detecting and selecting stimuli in my view.
However , the story is not complete even now.
Consider, the parallel path associated with response/ action.
One can be in various states of readiness to respond. If you are highly aroused and edgy you may jump at slightest stimuli or act with hyperactivity. On the other hand if you are aroused pretty less you may be sleepy/drowsy and sluggish in your responses. In the extreme case of sleep you wont respond at all. This arousal system as is well known is mediated by tonic NE levels. To me, this is the same alerting network – though this time attuned to responding and not detecting stimuli. And this is a non specific general readiness to respond.
So the first step to responding is being in various states of alertness as to whether and how fast you would respond to any stimuli in general.
The next step of course is based on cues you would be in state of readiness for a particular (set of ) action (s). The CS that leads to orienting response and attention to where the UCS may appear also primes a response set or UCR. To take our crude analogy, when you hear thunder you would be primed to seek shelter (or stay away from trees) etc). In a nutshell a response set will be activated and a goal needs to be maintained. This is the orienting network in action in the response leg. Note that all this is happening in parallel. That is Alerting stimulus leg and alerting response leg are running in parallel and so too are orienting stimulus leg and orienting response leg.
So the second step to responding is being primed with a particular response set. In ANT this would be responding with ether left hand or right hand .
And then the stimulus happens.
Given that the stimulus has happened a primary response gets activated, however there are competing responses : in the contrived laboratory settings this may be in the form of STROOP test where habitual response competes with more relevant response. So the third system kicks in suppressing irrelevant/ habitual responses that are not relevant. This as you would have guessed is the executive control – conflict system.
So the third step to responding is suppressing irrelevant responses. And this is beautifully accomplished by executive control – conflict system.
Finally, the ‘cold’ system needs to take over and finalize, initiate and execute the final action. This is the executive control- decsion network.
So the sequence is :
- Readiness to respond (Alerting)
- Priming for a particular response (Orienting)
- Stimulus appears
- Suppressing irrelevant responses (executive control – conflict)
- Selecting final response for deep processing (executive control – decision)
I should again emphasize that the stimulus leg and response leg run in parallel. I believe there is great value in talking about both legs when we focus on attention and underlying attention networks.
In my next post I will elaborate a bit more on underlying brain structures / functional networks and neurotransmitters underlying these networks .
PS: Some of my musings are based on studying these articles in depth and I express my debt to them.