In ‘Understanding Consciousness’, Velmans makes much of ‘representations’ which implies building models of the outside world within the brain but I’m wary of this notion. It is in contrast to Rodney Brooks’s idea that ‘the world is its own model’ which says that there is no need to have any ‘representation’ if you can just process the input sense-data (input processing = classification = taxonomy). The brain can minimize the amount of processing needing to be done by taking best advantage of the resources that already exist.
This is not like, say, a man-made CMOS image sensor in a mobile phone. For this, an image snapshot is taken and data is read out of the sensor and stored within DRAM memory before the processor ‘sees’ the picture and maybe runs Bayer transforms on it and performs JPEG compression. Brooks’s idea becomes obvious with hindsight: biology is not constrained by different materials that affects electronics and can come up with more efficient solutions. Why require (i.e. evolve) additional hardware to store images and at the same time ignore the readily available input sense-data for more than 99% of the time?
Velman’s discusses Ben Libet’s experiments (in 1979, not the more famous in 1985) of contrasting tactile sensation with direct microelectrode stimulation of the somatosensory cortex. Tactile stimuli can be applied 100ms after direct stimulation of the cortex but consciously be perceived as starting 100ms beforehand! Paraphrasing Libet’s explanations:
“the brain records the actual time of arrival of the stimulus at the cortical surface. The brain then enters this into the representations of input that it constructs.” (p. 234).
“a skin stimulus produces a … potential on arrival at the cortical surface that acts as a ‘time marker’ for its time of arrival, and the brain subjectively refers experienced time of arrival ‘backwards in time'” (p. 235).
This ‘subjective referral’ is ‘backwards in time’ to the order of about 200ms.
But there is no need for this elaborate explanation, which seems to owe more to computer programming than biology. Considering the brain as a massive bundle of neurons, there will be a massive number asynchronous (chaotic) feedback loops. There is strong evolutionary pressure for adaptation that minimizes the delay from input to action. “I see a lion. Run!”. Antelopes that react within 300ms will survive over those that take 400ms to react. Over time, this will favour a reorganization towards minimal neural connections in a forward path with more connections growing in the feedback path in order to achieve the same function with a shorter initial propagation delay. Direct stimulus of the somatosensory cortex is much more likely to tap into the fed-back path rather than the forward path and so will take longer to become consciously noticed than a tactile stimuli that feeds directly into the start of the forward path. Could this not account for the 200ms negative time?
This theory is also consistent with Libet’s suggestion (see p. 252) that there is an unconscious (preconscious) determination of action from input with consciousness serving to monitor this action and, if necessary, to veto it. As a result, consciousness is always seeing things ‘after the event’. There is an unconscious ‘reflex’ action followed by possible executive veto. Velmans puts it very nicely: maybe there is no such thing as ‘free will’ but there is ‘free won’t’!
This is analogous to how a manager might run his team. If he insists on approving everything his subordinates do before it ‘goes out of the door’, the delay in processing work will be considerable compared with if he empowers his subordinates to act but monitors what goes on. In this analogy, the manager is always reacting to circumstances, seeing things after the event, which is being in control – of sorts.
Velmans notes an asymmetry in this behaviour: why isn’t there preconscious processing for the veto function? But again, this makes sense from an evolutionary point of view – there is no need for any hardware to evolve to ‘rapidly not do something’ in the way that there is for a need to ‘rapidly do something’. Abortive false alarms are better than being eaten alive.