You'll feel great for doing the work yourself, though it might not save you much if you have high value assigned to your own time. To me, it's well worth doing yourself. In reality you are just unscrewing one 22mm bolt that just happens to be called an "O2 sensor. Removing The Airbox removing the inlet tubes from the air box:. Here is a list of what you need to check: Gas cap. Fuel rail connections, especially if it is a If they are over tightened they can crack.
Connections on the fuel filters. Make sure that the rubber isn't cracked or leaking. Verify that both 31 see below hose clamps are sufficiently tight but not choking the hose. Same deal, make sure the rubber isn't cracked or stopped up. Verify that hose clamp 35 see above is sufficiently tight but not choking the hose. Thanks, ernie. To check the connections that Ernie lists above, use your finger. Rub your finger under and around each fuel line connection e.
When your finger smells like fuel, then you've found at least one leak. Wipe the entire area concerned with yet another it's important not to re-use a rag here so as to avoid tainting the smell rag, and dispose the rag so as not to confuse yourself with fuel smells. If you don't have a fire extinguisher rated ABC, then get one before you next start the car.
Likewise, turn the ignition key to the Run, not to the Start, position to see if you can find any more leaks after your yellow fuel-safe teflon-tape repair above. Once all of your connections have no fuel smell with the ignition key in the Run position, then start your car and recheck all connections again with your finger while the engine is running. Try this simple thing if you still smell fuel I had a gas smell develop on my Spider which the dealer could not solve even after inspecting all hoses and replacing the flapper valve near the top of the system at considerable expense.
Thanks, VinnyCar. I smelled fuel in my when I first got it especially when the tank was full or when I did corners. Took a lot of trouble tracing the trouble. In the end, it was an "overflow escape valve" not sure of the correct name near the fuel cap that was stuck and needed to be cleared. Thanks, SFChallenge.
CF does point out, however, that the more open cloth-type performance air filters e. OEM paper filters don't vent gas fumes, however. A bad spark plug, spark plug wire, coil or coil connection , bad catalytic converter, or bad O2 sensor can also leave a detectable smell of fuel in the vicinity of the catalytic converter area. Unique Fire-Prone Areas. Unique Fuel Smell Areas. Unique Testarossa Fuel Smell Areas. Unique Mondial Fire-Prone Areas. The major electrical issues on most Ferraris are: 1.
Failing car battery 4. Failing alternator system 5. Corroded or burned or otherwise failing relays 6. Reverse current that is traveling backwards through dash bulbs that have burned out Use Stabilant 22a for all of your electrical your connections. Insure that you've got an adequate ground strap for your car battery and for your alternator. Insure that you have a decent car battery e.
Insure that your alternator is giving enough amps in its output under load. It's fairly common for a parked to drain the battery over a period of time of as little as two weeks, but it doesn't have to be that way.
It's relatively quick and painless to track down the usual electrical culprits. First , have your favorite auto shop install an ammeter amp-meter inline with your battery and easily viewable or if you know what you are doing, you can buy an ampmeter from RadioShack and install it yourself.
Again, note the current draw. At this point you should know which half of your circuits are drawing the most current when your car is supposed to be "off. By following this process you will quickly identify your problem circuit s , showing you which part of your needs anti-corrosion fluid, wire-brushing, Stabilant 22a , and perhaps even new components. It seems as our cars age, the construction of the original relay and fusebox allows moisture and trollage to come into the relay.
To illustrate the point, the inside of your relay can be estimated by the condition of metal on the outside via the surface deterioration. Thanks, Russ Turner. There seems to be a general consensus as a good idea for replacing the two old relays for the left and right fuel pumps found in the passenger footwell as well as replacing the two old relays for the left and right fuel injectors with the newer and cheaper!
They are cheap at RockAuto. Breakers that can replace fuses. Bosch relays as used on Ferraris and various other makes of cars number their metal pins with a standard category system. Here's what you do to test the oil cooler fan - very simple and this procedure is valid for using on other Bosch relays such as the fuel pump relays, see chart above. Pull out fuse It's the third one see picture above from the right in the lower row 20 amps.
Make sure it isn't blown, and put it back in. Pull out relay "O". It's the center one in the bottom row in the footwell panel. There are five slots in the relay panel for the five terminals on relay "O". Take a jumper wire with two male spade terminals and jump the two positions shown in the picture top center to bottom center, terminal 30 to terminal It doesn't matter whether the ignition key is on or off.
As soon as these terminals are connected, the oil cooler fan should come on. If it doesn't, then there is a problem either with the wiring, the fan, or the ground. If it DOES come on by jumping the terminals, but never comes on any other time, then we can look at some other possibilities.
The thermoswitch for the oil cooler fan should close at 90 degrees C, turning on the fan, and open again at 80 degrees C, turning off the fan. There is normally no power running to the thermoswitch or to the fan motor, so don't check for power in the back.
Then you need to adjust the window guide on the door. It is the little black irregular oblong part at the side edge of each door. When you open the door you will see two black allen screws above and to the side of the red door light. Loosen those when the window is down, then raise the window. Now tighten the two allen screws back down. Alternators and Voltage Regulators. When driving, this light may come on to indicate that the alternator has died and is no longer charging your battery.
You can still drive your for a few miles, but soon you will be stalled on the side of the road if you take no positive action. You were warned! That being said, your may also warn you of a failed alternator before you even leave your garage. However, with a failed alternator, at least in some cases your battery light won't illuminate for that light-test prior to starting your car.
Again, consider yourself warned if you don't see your battery light prior to starting Instead of the normal alternator warning via the battery light coming on while driving, if you don't see the battery light prior to starting, then you won't see it while driving even if the alternator has failed. Thanks, Bee Jay! Complete pictorial step-by-step alternator removal, rebuilding, and replacement Ernie rocks!
Ernie rocks! For the rebuilding of your alternator models use two different alternators based on year - Delco and NipponDenso , you might consider using some copper heat transfer compound between the rectifier heat sink and housing for better heat transfer and electrical contact. First, disconnect the battery. That's important. Then you loosen up the belt tensioner take note of the belt tension so you can re-tension later , remove 2 bolts on the alternator disconnect the wires on the alternator and down she drops.
F and Alternator Issues. Mondial T, , and F Alternator Replacement. The Delco part number is The newer, cooler-running "iceberg" version is here. That unit may require the universal braket adapters sold on the same site. I already had one alternator melt down, replaced it along with the battery. Now the tell-tale battery light is coming on again and the tach is going nuts at higher RPM - the signs of impending alternator doom.
I just went through the archives and it looks like Elreg has the parts for the Delco. Other than that, a direct bolt-in. For 's and F's, the obvious first place to check is the "battery cut off" AKA "battery disconnect" switch.
Both ground cables can be attached to terminals on the same side of this cut off switch in order to bypass it in case the cut off switch is making a bad electrical connection. After bypassing the cut off switch, check the heavy cables at both the junction box in the rear engine bay and under the dash in the U. S passenger foot box? Found problems here numerous times. Thanks, davehelms. You will need this out so you can access the four bolts that hold on the front. These have to come off so that you can unplug the side maker lights, and to loosen the four bolts, two on each side, that hold the bumper in place on the sides.
This will give you a little bit more room to get your hand up under the bumper to disconect the plugs for the fog lights and turn signals. You will see two L-shaped rubber pieces, on the back of the bumper, with some metal spacers on the bolts that you just slid out.
You can take those off, so they don't get lost at the shop. A few tips. After you have the bumper off, while it is still fresh in your mind, put the screws and bolts back in the holes and on the bolts they came from. This way you don't loose any. Also when you go to put the rubber pieces back on, the L goes on upside down and the bottom of the L points out, toward the side of the car.
Rear Bumper Removal. Steel can be inexpensively welded, and additional steel plating can even be added if desired at the two major in rear weak points. Note the black metal plate that has been welded onto the Ferrari sub-frame in the center of the picture. This is repeated on all four sides of the frame rail that had the crack. By welding this "box" around and to the cracked OEM rail, the frame gains strength You can also go to a body shop supplier , it is 3m body seam sealer.
Thanks, MDShore This cable must be manually pressed back in so that no cable is exposed see picture. When you pull out the handle in the engine compartment, it doesn't retract all the way by itself, you have to push it fully home. If it's part way out, the plunger does not extend all the way to keep the door shut, even though the electrical mechanism is working fine. It won't work properly if the handle is part way out like this - push it in!
GeorgW reports that mechanical re-adjusting of the door itself and the plastic plate which connects to the plunger can also solve the problem. Outer Mirror Disassembly Thanks, Miltonian! I bought a mirror from Kragen I picked one that was thinner than the original ; used Goof-Off the solvent a lot of guys on here use for the sticky parts cleaning and a trim removal tool to gently pry off the original mirror.
I pryed at the original mirror with the trim removal tool as I worked Goof-Off into the back of the OEM mounting plate. As wingfeather found out a good yank removes it. Then I cleaned the area with an alcohol wipe to remove any remaining residue. For the life of me I can't recall what bonding agent I used, it is the stuff they sell at Kragen right next to the mirrors.
That allowed me to place the mirror quite a ways up the windshield to greatly improve visibility. The most important thing to have on hand is a six-pack of beer to calm your nerves so you don't think about how much it will cost to replace the windshield if you muck it up Part 53 above white plastic often breaks.
It's a common plastic automotive cable retaining clip same as for all Chrysler cable latches as well as Ferrari , F, and It's Ferrari part Pep Boys calls it a "door lock rod clip.
You can open stuck doors on a from the outside when the window is down, using a long slotted screwdriver. Look down inside the door, at the rear, in the area of the door latch, towards the outer side of the door. If you look carefully, you can see where the cable from the inner door release handle runs straight upward into the latch mechanism. If that cable is attached correctly on each end chances are that it isn't , then when you pull UP on the inner handle, the cable pulls DOWN on the lever on the latch.
By reaching inside with the screwdriver, you can push straight downward on the same lever, and the door will pop right open. I just did it on my own to be sure. Once you get the door open, you will need to remove the panels to find out why it isn't opening correctly. Chances are that the linkage has just come unattached.
Open the door 2. Disconnect the wiring buckle from the door jamb just to allow room to work 3. Using a punch, pound the exisiting pin out, from the bottom up. It takes a little time, but just be sure not to hit any paintwork! Once the pin is basically out, use a needle nose pliers to remove it entirely.
My old pin looked nothing like the new one: old pin was probably a single wrap of steel and fell apart as I needle-nosed it out. New pin was several wraps of steel, and much more substantial. From the top, insert new pin. Using a punch tool and a carefully swung hammer, drive the new pin into the hinge.
No more popping sound! Part number , and it's fixed. NOTE: the F diagram at right below uses a screw 8 rather than the 's pin. This screw 8 and nut 43 may be a desirable upgrade for owners. I've removed the door grates today and there are no screws under the door handle.
Thus, if one cable breaks, you can still access your front luggage or mid engine. Each pull cable opens 1 front latch, so both reserve cables must be pulled to access your luggage compartment. You push one side down with one hand and pull the lever with the other.
And have some one push down the other side at the same time I still have the old Stabillis hood strut, so I took it to the Audi spare shop again and bought another one. Both ball joint ends including the shaft are chrome while the original are black. The body Cylinder is black.
Be warned that it is slightly fatter than OEM and extra strong even with just one. If you change both, you might bend the bonnet trying to close it!
Even with just one on the rear, I have to put a hand on the cover before I pull the level as it will spring up. First, prop open the bonnet with a long rod. This operation entails removing the support struts, after all. There are the main two styles of struts. You'll have one or both if a prior owner mixed types on your car. One has a thin wire "lock" that holds that round ball inside the black plastic cap on the end of the strut.
The very end of this wire "lock" is just a straight little tip of metal, and it pokes into the black plastic cap on the strut to lock the ball so that the strut stays installed.
If you have this type of strut, then place a flat-blade screwdriver as shown in the picture in the link above and pull the little metal lock tip out of its hole in the black plastic cap. Then put your hand on the top of the strut and yank. One end of the strut is off. Repeat as needed. The second type of strut has a wider wire clamp.
Easy peasy. There are two common faults, the locks both cycling up and down, as well as the locks going into the lock position briefly and then popping back up into the unlocked position i.
The latter is typically encountered when one or both doors is not fully closed. This is a feature of the , not an actual "fault. If you open a door, push the lock on the door manually and then close the door, this seems to make the lock motor do it's up and down cycling dance at least when the lock hasn't moved all the way into the locked position. In this case unlocking the doors with the key instead of the center console button should stop the locks' dance. Thanks, porkface.
Alternatively, it may be due to adjustments because if I grab the lock button on the door and yank it upwards on the upstroke so to speak, it will suddenly stay in the unlocked position instead of cycling Thanks, jkuk. My did it.
Turned out that the door knob was screwed too far into the door not allowing enough movement on the relay switch which made the relay go nuts.
BTW: the panel was off today because the key would turn, but the lock wouldn't move. There's a small pin in the end of the lock that fell out so turning the key turned the tumblers, but nothing else. I suspect the pin went out the door's drainage hole. Until I find a used lock a locksmith couldn't help , I used a piece of Romex ground wire cut down to the proper size.
It was malleable enough to do the job. Thanks, clydeM. A broken wire or bad connection in the door connector very common can cause this. It was exactly this fault plus a crackling speaker that made me re-wire it. Thanks, Chrisx I had almost the same problem, started when the wife slammed the door, can't blame her, door seems heavy and people don't realise how easy it closes.
The unlock signal kept repeating because it thought the door was still locked,[I think] and it was. As I remember, I had to turn off the battery to get the "psychotic" action to stop, drove home and all was well until I tried to unlock the doors again, then same thing over and over.
I had to disassemble the door from the outside to get to the lock and fiddle with the linkage btw the lock and solenoid. Thanks, Markie TADA, fixed. When I turn the key In the drivers side door lock it would just turn without locking the door. I figured it could be tumbler problem. So today I removed the inside door panel. The "C" clip in the back of the mechanism had popped off and there was a spring and a few other parts bouncing around.
Thanks, amorepresto. If one door isn't fully closed, the locks will go down for you and then pop back up. Ferrari calls it the Monica feature. Ba dum bump. It might also be a faulty lock-logic unit 3 in diagram below Ferrari pn: under your passenger fuse panel.
You may have a broken plastic linkage piece on the back of the lock cylinder on the driver's door. This would explain why there is no resistance when you turn the key, and it is a fairly common problem. Be prepared for a shock when you go to purchase this piece. The inner door panel can be removed with basic tools, again, there are good instructions here.
Or the outer door panel the "strakes" can also be removed in just a few minutes, and actually gives better access IIRC. It's the little red piece that breaks. It's also at least possible that the linkage arm has simply popped out of its socket. You need to remove the door panel and take a look. Total cost 19 cents and 8 hours my labor. See diagram above 31 second from the left. Thanks, AmorePresto. Picture thanks to JJStecher. The center section has two long studs on the back that go through the firewall and have nuts in the engine compartment.
Look between the plastic coolant pipes, a couple of inches outboard of the struts for the engine lid. Once you get the nuts off, you can slide the center panel forward, then I think you can get at the screws to take off the side pieces. Interior color can be tricky, I ordered Saddle from Leatherique using the GAHH color chart and the dye was lighter than the tan interior on my There is another Saddle color called E II which is slightly darker.
Anyway, in the end I sent them an sample cut out from the seat to make sure I get the right one. If you are replacing everything may not be a big deal though Thanks, Jagbuff. Custom Plastic "glass" Fabricator. Replacement Windshields and Door Window glass: info glastec-gmbh. Email Windshield Price Quotes. Slow Window Aftermarket Fix. Bosch Window Motor Repair Pictorial. Here are the steps. Then remove the covers under the gas tank, and the shift linkage. For the you will also need to remove the under tray and diffuser.
DO NOT use the one on the bottom of the engine. They are in the middle, upper, front part of the engine. It is on the left hand side of the engine bay. If you have access to a shop lift this step isn't nessesary. You will then need to remove the heat sheilds.
Now remove the 4 nuts, two on each side of the bumper, located near the wheel well. Then remove the 4 nuts that hold the bumper to the frame, they are by the muffler, and remove the bumper. The kind that has four caster wheels on the bottom and carpet on the sides. But don't remove the nut that holds the shock in place on the subframe. For 's you will need diconnect the ride adjustment motors.
As you are raising the body feed the ECU cables through the holes. Doing it this way makes it much easier. After you have the body off of the engine you will need to rotate the engine 90 degrees in order for the shock towers to clear the bottom of the cross bar that the heat sheild is connected to.
If you are going to use the two floor jack method, you will need to make sure that the floorjacks have a lift height of at least 20"". Other wise you will not be able to get the car high enough to roll the engine out.
Also make sure the you are using jack stands and wheel stops. Happy wrenching. After I had everything drained and disconnected it then took me another 4 hours to remove the engine, and then about 5 more hours to put it back in.
Alot of that is double checking your triple checks. Remove the engine lid. Working inside the car, slide the seat forward. Remove the leather covered panels on the bulkhead,both of them. Underneith you will find the ECU's for the engine. Unbolt the ECU's from the bulkhead, disconnect the computer and unplug the yellow and white connector. The rest of the pull is fairly basic. Remove air box assembly, and exhaust heat sheild.
Drain your fluids, disconnect hoses, throttle cable, vent lines from the oil tank, clutch line, brake line, parking brake cable, there are 2 small vac lines and 2 electrical connections that must also be removed on the left side of the engine bay toward the front of the engine.
They are for the e. Oil lines must also be disconnected from the oil tank and oil cooler. Undo the exhaust from the top mounting flanges.. Remove all the inner fender sheilds, disconnect the ABS connectors white and remove the overflow tube from the expansion tank which is clamped to the shock tower.
If you are pulling this engine on the floor.. Undo the 4 nuts on the shock towers, and I sometimes even remove the nut on the top of the shock along with the washer. This makes it alot easier to remove when the body flexes inwards during remove of the subframe. Undo all the other mounting points 12 bolts at the front of the car 6 on each side and 4 at the rear, once you have the subframe supported. The ones at the rear you will see when you remove the inner fender panels.
One bolt hole is right beside the cat horizontaly And the other is right above it verticly. When you start to lower the engine out.. The body will flex almost a full inch inward during the removal. I use a turnbuckle brace inbetween the shock towers to keep this from happening. It makes the reinstall very easy.. As you are lowering.. Do this by lowering the engine about 3 inches at a time, and working inside the car push the harness towrds the engine. The hole they have in the firewall is a little difficult to work with because it is small and the harness connectors barly fit through the hole.
They must be put through one at a time. Hope this helps.. Aluminum gear teeth on SKF bearing internals are sold here: Forza. Engine Out Service Pictures thanks, ericzz.
Yes you do need the water pump changed. Make sure that they also change out the thermostat. There is a clip on the top of the thermostat that has a week spot. The clip is held in place by two ears that are pinched in place.
Over time these can break, and the clip will go into one of the radiators. I know this first hand. It is a pain to get out because you don't know what radiator it is in. When it happened to me I found it in the radiator just above the oil cooler. Make sure that they change the straight hose that runs from the water pump to the cross over pipe, that is under the intake plenum.
Have them give you a copy of the measurements of the valve clearances, before and after they adjust them. You want this so that they don't charge you for adjusting 32 valves when, let's say, only 9 valves really needed adjusting. Also: Keep your ALL of your old parts. The old seals, old valve shims, tensioner bearings, and especially the water pump. To take out the engine yourself using two floor jacks you have to remove the rear bumper. First, just go to Home Depot, or any hardware store, and get a pound moving dolly.
They are flat with four casters, the wheels are solid rubber, and two of the sides are wrapped in carpet. Then all I did was jack up the rear and center the dolly under the subframe so that the frame sat on the carpeted sides.
The big ones, not the little cross beams, same for the dolly. Drain the fluids oil, coolant. Remove the 8 bolts that hold on the rear bumper. Next I removed the rear wheels and splash guards and sat the car down on the dolly. I didn't use anything to hold it in place. The engine is plenty heavy and the weight keeps it still. The dolly's are so cheap, you could get two of them if you want to feel better. I used just one, and it held up with no worries.
Total Time: 4 hours. ShanB used multi-colored zip ties to help him remember where everything connects upon replacement He used Ernie's two jacks method above and a little care. The car really does not need to be raised very high at all and the engine rolls out quite nicely once everything is disconnected.
His Total Time: 6 hours. Coat all electrical connections on engine with Stabilant 22a , too. FatBillyBob says that you don't have to roll the engine out; that you can change the cam belt by simply turning the engine 35 degrees while on the dolly. Here's a tip: So far the only challenge has been removing the idler bearing without damaging either the aluminum shield behind it or the pulley bracket.
The problem is that there's very little room behind the bearing and most puller jaws simply won't fit. I read Ernie's tip about using a socket to leverage against and found a good quality two jaw puller with thin tips that fit perfectly. I used a deep socket with a washer on top as internal leverage for the puller. Thanks, ShanB. Improved 30K Service Parts Package. The amount of valve shims can only be determined after clearances have been checked.
If you car is equiped with a charcoal canister, it will need to be changed if the car has over 50, miles on it. You will also need oil, water, coolant, brake fluid, and gear box oil as these are flushed out and replace with new fluids. Intake Removal thanks, ernie , Miltonian , and others. Replies 4K Views 1M. Today at PM karnage Grexeno Dec 16, Sticky 2, Replies 35 Views 3K. Yesterday at AM ioriyagami. Our gaming hardware is linked to slave labour, concentration camps and genocide.
Here's what to do about it. Replies Views K. Dec 28, Messofanego. What is your opinion of Banjo-Tooie? Replies 21 Views A moment ago Mary Celeste. Arta Today at PM 2 3 3, Replies Views 3K. A moment ago Arta. Replies Views 7K. A moment ago Antipode. Kalentan Friday at AM 10 11 12 13 14 20, Replies Views 21K. A moment ago Wanderer5. Lunchbox 21 minutes ago 97 2. Replies 2 Views A moment ago RetroRunner. News Loudninja Today at PM 3 4 5 6 7 12, Replies Views 13K. A moment ago benzy. The art direction of the Pokemon Let's Go games is the peak of the entire series.
MasterYoshi Today at PM Replies 48 Views Replies 74 Views Replies 72 Views 2K. Replies 3K Views 48K. Microsoft - Entertainment - OT. Doctrine Dark Today at PM 6.
Replies 6 Views Replies Views 19K. Zeta Ori Dec 15, 2 3 4 5 7, Replies Views 8K. Steam Deck - January Update. Replies Views 5K.
News - Valve. The system can only be aware of the knowledge reported by certain mental operations and those in the input buffers, and even this knowledge does not necessarily get enough attention to reveal its implications.
Although the system only has limited self-knowledge, it nevertheless make self-control possible. A widely agreed conclusion in psychology is that a mental process can be either automatic implicit, unconscious or controlled explicit, conscious , with respect to the system itself.
Such models are also needed in AI, even though the purpose here is not to simulate the human mind in all details, but to benefits from the advantages of both.
In general, controlled processes are more flexible and adaptive, while automatic processes are more efficient and reliable. Such a model often uses meta-level processes to regulate object-level processes Cox, ; Marshall, ; Shapiro and Bona, ; Rosenbloom et al.
This meta-level deliberative control does not change the underlying automatic routines, but supplement and adjust them. This design is different from the metacognition implemented in the other systems Cox, in that the operations in NARS are light weight and can be accomplished within a constant time, rather than decision-making procedures that compare the possible actions in detail with a high computational cost.
Furthermore, the system does not merely remember whatever it has experienced, but selectively keeps them, and generates conclusions and concepts to summarize and generalize the experience, so as to deal with new situations efficiently. Instead, the concepts and statements capture the regularities and invariants in its experience, so are fundamentally from the view point of the system itself.
A special type of knowledge is the skills , i. As described previously, each operation is evoked when a certain condition is satisfied, and compound operations can be formed.
NARS has the ability of self-programming, in the sense that the system can organize its atomic operations into compound operations recursively and use them as a whole, so as to avoid repeated planning or searching Wang, b. The motivational structure of the system is under constant adjustments and developments and is not fully specified by its designer or users. NARS is built to accept any task expressible in Narsese in any time, although the priority of each task will be adjusted by the system, and the system may even ignore some given tasks, as the consequences of conflict resolution, preemptive action, redundancy reduction, etc.
Since all aspects of the object-level can be learned, everything expressible in Narsese is learnable, in the sense that it can be entered into the system, derived by the inference rules, as well as modified by new experience. Consequently, NARS is more sensitive to its experience than most AI systems developed so far, and learning happens in several different forms in various parts of the system.
This treatment of learning is fundamentally different from the current machine learning paradigm Russell and Norvig, ; Flach, , since in NARS the learning processes do not follow algorithms and nor do they necessarily produce problem-specific mappings Wang and Li, This sensitivity to experience does not mean pure subjective or arbitrary behaviors.
The objectivity in knowledge comes from communication and socialization. On the other hand, in NARS, the meta-level knowledge is built into the system and immune to experience-triggered modification. This level includes the grammar rules of Narsese, the inference rules of NAL, the basic routines of memory management and inference control, the set of mental operators, etc.
Even taken self-awareness and self-control into consideration, this built-in core is still fixed. We consider such assumptions unrealistic and therefore is irrelevant to the design and development of AGI systems.
This topic can be addressed from many different perspectives Blackmore, , and there is still less consensus on its basic form and function. Many people consider it impossible in AI, although there have been attempts to produce consciousness in computers Baars and Franklin, or robots Chella et al. Here, we focus on the so-called hard problem, that is, how physical processes in the brain give rise to subjective experience Chalmers, First, a sensation is produced by a sensor; then, a perception depends on the generalization and association capability and the available concepts of the observer; finally, when the perception eventually becomes a description, the system must have paid enough attention to it, which in turn demands a relevant motivation, a proper emotional status, and so on.
Beside the culture heritage, our descriptions of the world heavily depend on the common sensorimotor mechanism of the human species, which is not necessarily shared by all cognitive systems, like the other animals or robots, either the existing ones or the future ones. This conclusion is not limited to robots. Actually EGS can be applied to any system, as far as it has interaction without its environment. A direct implication of the above conclusion is that intelligent systems in the same world may form different descriptions of the world, due to their different sensorimotor organs, concept repositories, motivational orientations, etc.
In this situation, all these descriptions are valid, even when they are incommensurable. The same is true within the same system. If the system applies two different sets of sensorimotor mechanisms to the same process, it may get two descriptions, which are correlated, but incommensurable, and cannot be reduced into each other.
As described above, NARS has internal experience about what is going on inside the system, which directly comes from the mental operations and the related introspective functions.
When the system also learns how its own design works from a third-person perspective, even when it is given a way to observe its own running process at the machine language level, it will also have two incommensurable descriptions with a gap in between. In summary, we believe that the design of NARS enables the system to have consciousness, and the related phenomena can be explained without being reduced into phenomena in neuroscience Koch, or quantum physics Penrose, In AGI systems, although initially the conscious functions will be relatively simple and poor, they will become more and more complicated and rich, as the research progresses.
The fact that we cannot directly sense them cannot be used to deny their experience, just like one cannot deny the consciousness of another person simply because one cannot directly know what it is like to be that person. Self-awareness and self-control are important cognitive functions needed by advanced AGI systems Chella and Manzotti, For a system to solve various types of problems, especially novel ones, it needs to know about itself, as well as to adjust its own working processes, so as to efficiently produce the best answer it can find with the current evidence and resource supply.
An AGI system can learn how itself works using its introspective capability, especially the mental operations. These functions enable the system to better adapt to its environment and to carry out its various tasks more efficiently. Even so, it can never fully know itself nor can it have complete self-control.
Although the study of self-awareness and self-control in NARS is still at an early stage, the conceptual design described above has been implemented, and is under testing and tuning.
There are many details to be refined, and many self-related issues to be further explored, like those discussed in the studies by Hofstadter and Blackmore We believe the overall design is in agreement with the scientific knowledge on these processes in the human mind and also meets the needs and restrictions in AGI systems.
We also believe that almost all self-related functions observed in the human mind will be reproduced in AGI systems in principle although the details will be different. Furthermore, these functions should not be modeled one by one in isolation, but all together according to the same basic principles of intelligence.
PW proposed the overall structure and drafted Sections 1, 4, and 5. XL drafted Section 2. PH drafted Section 3. All authors revised the whole article. The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Thanks to the anonymous reviewers for their helpful comments and suggestions. Thanks to the other members of our research team for proofreading the manuscript.
Arbib, M. Emotions: from brain to robot. Trends Cogn. Baars, B. Consciousness is computational: the LIDA model of global workspace theory. Bach, J. Oxford: Oxford University Press. Google Scholar.
Bieger, J. New York City: Doctoral Consortium. Blackmore, S. Consciousness: An Introduction. Chalmers, D. New York: Oxford University Press. Chella, A. A cognitive architecture for robot self-consciousness. Wang, and B. Goertzel Paris: Atlantis Press , — Cooper, R.
Cognitive control: componential or emergent? Cox, M. Metacognition in computation: a selected research review. Flach, P. Franklin, S. Goertzel, and P. Freud, S. The Interpretation of Dreams. New York: Avon Books. Translated by James Strachey from the edition.
Gibson, J. Gillespie, A. Intersubjectivity: towards a dialogical analysis. Theory Soc. Goertzel, B. Hofstadter, D. New York: Basic Books. Kahneman, D. Thinking, Fast and Slow. New York: Farrar, Straus and Giroux. Koch, C. Greenwood Village: Roberts and Company. Kowalski, R. Logic for Problem Solving. Luger, G. Boston: Pearson. Marshall, J. A self-watching model of analogy-making and perception. McCarthy, J. Minsky, M. The Society of Mind. New York: Simon and Schuster. Nagel, T. What is it like to be a bat?
Peirce, C. Penrose, R. Oxford University Press. Piaget, J. The Origins of Intelligence in Children. New York: W. Translated by M. Poole, D. Cambridge: Cambridge University Press.
0コメント