Graphical password systems are said to have been a disappointment. It shares the same drawbacks with textual passwords to a large extent, i.e., easy to forget if difficult to break, easy to break if easy to remember and vulnerable to shoulder surfing. The merits over textual passwords, if any, are quite marginal.
I believe we have overcome major weak points of the conventional graphical passwords (1) by making a good use of episodic-memory, (2) by counting the sort of errors that the legitimate user can make separately from the sort of errors that the legitimate user is unlikely to make, (3) by enabling the user to reduce the size and location of the pictures to help prevent shoulder surfing.
(1) The logic is quite simple: Most of us, even when caught in panic, can easily find several meaningful pictures colored by their own episodic memory from among similarly looking but meaningless decoy pictures. When you have found those unforgettable pictures that you had registered, you have proven your identity while rejecting third persons who have no clue to your old personal memory. (Patent granted in Japan and pending in USA for this algorithm)
Textual passwords of episodic memory are known to be too weak, not the least because episodic memory that can be presented in texts are objective enough to be easily guessed by third parties who are in a position to know your personal history. But episodic memory that can only be presented by images are not.
(2) We can make errors in placing the cursor or finger on target pictures when we have correctly recognized them, which is a typical example of the sort of errors that the legitimate user often makes. By separating such errors from the sort of errors that the legitimate user is unlikely to make, we can offer the identity authentication that is easy and less stressful for the legitimate user but difficult and more stressful for the attacker. (Patent granted in Japan and pending in USA for the algorithm)
(3) Attackers can now resort to a tiny video camera for shoulder surfing. With its function of slow-motion replay, it would be no use of moving your fingers very quickly above the keyboard when typing textual passwords. Pictures, when reduced in size and moved in location, can be easily hidden from video cameras to say nothing of naked eyes.
For further details, refer to “Outline of Mnemonic Security” shown at
Below is some more related information.
A. Fundamental Requirement of Democratic Society
Every citizen in democratic societies must be given the universally practicable tools to verify their identity, whoever, whenever, wherever.
They did have such tools in the millennia-long pre-digital societies in the forms of hand-written signature. Signature can by no means be performed by an unconscious or dead person, and irrespective of whether rich or not, whether educated or not, all the citizens who could live a social life could verify their identity with the signature. The digital technologies, however, are making the signature invalid, whereas contemporary citizens are not given a practicable alternative as yet.
Mnemonic Guard could well be the top-runner of the group of solutions that can be the digital alternative to the conventional signature.
B. Episodic Memory (Autobiographic Memory)
We emphasize the merits of making use of long-term memory, visualized episodic memory in particular. Below is about what the episodic memory is.
What is Episodic Memory?
http://www.wisegeek.com/what-is-episodic-memory.htm
Episodic memory is a form of memory which allows someone to recall events of personal importance. Together with semantic memory, it makes up the declarative section of the long term memory, the part of memory concerned with facts and information, sort of like an encyclopedia in the brain. The other type of long term memory is procedural memory, which is the how-to section of the brain.
The primary contrast between episodic and semantic memory is that episodic memories are memories which can be explicitly described and stated, while semantic memory is concerned with concepts and ideas. For example, the concept of a table is housed in the semantic memory, but when someone describes his or her kitchen table, this is an episodic memory. Procedural memory can also interact with declarative memory, as for example when someone drives a car, using procedural memory to remember how to drive, semantic memory to define a car, and episodic memory to recall specific driving experiences.
Episodic memories can pertain to general or specific events, such as what it feels like to ride a train, or a specific event which occurred on a train. It can also include facts, such as the names of world leaders, and so-called “flashbulb” memories, which are formed during periods of intense emotion. A classic example of a flashbulb memory from the 20th century is the assassination of President Kennedy, an event which was vividly remembered by people who were alive at the time.
It only takes one exposure to form an episodic memory, which is probably something which evolved early in human evolution, to teach people to avoid making potentially deadly mistakes. For example, someone who almost drowns as a child will often develop a fear of water in response to this single experience. People engage in episodic learning every day, but children often provide very striking examples of episodic learning, since they are exploring a world which is primarily unfamiliar to them, and hence they constantly have new experiences which are filed away in the episodic memory.
This area of the long term memory is a critical part of identity. People are shaped by the events they participate in and interact with, and loss of episodic memories can cause people to experience confusion or distress, as they lack a context for their identities. Some researchers
have suggested that episodic memory sometimes turns into semantic memory over time, with the brain lumping a family of similar experiences together to create a semantic concept. For example, distinct memories of various burns may be bundled together into the semantic memory to provide a concept of “hot,” along with information about which kinds of things tend to be hot.
C. Widest Choice
People who first hear about Mnemonic Guard might well suppose that Mnemonic Guard is exclusive of character-based passwords and that they would be forced to use only photos or graphics. The real picture is that users of Mnemonic Guard will be free to choose whatever they want to use or whatever they are good at from among photos, graphical symbols and characters/numbers, say, whatever is visual. It is as though we provide escalators and elevators along with staircases so that people could choose any of them depending on their physical conditions.
D. Office Environment versus Mobile Environment
What is effective in the unguarded and unmonitored mobile/field environment is also effective in the guarded and monitored office/home environment, but the reverse cannot be said. Mnemonic Guard is certainly practicable and effective in mobile environment.
Two-factor authentications are no doubt very effective in the indoor environment, but not so much in the outdoor where the combination of the mobile device and the token can be found at 100% probability on/around the clothes or bags of the person who carries them around . Can we expect thieves who can pick up the device will always forget to pick up the token?
Biometric devices with the tradeoff of FAR and FRR come with rescue passwords, which the users of the devices have to manage by themselves when moving around in the outdoor environment. This does not make a solution to the problem of weak passwords.
What we can really rely on is the memory-based identity authentication in the outdoor environment.
E. Cost Saving Effect: The cost of re-issuing a password per person at security-sensitive corporations is said to be well above US$100 including lost productivity.
Users of Mnemonic Guard could choose what to take up from among characters, symbolic graphics or photos depending on their skills, tastes and memory capability, which means that the cost could be drastically reduced.
F. Trend: Microsoft is reported to be coming up with a pictorial password system for its upcoming Windows 8, with which we can verify our identity by correctly tapping or moving our fingers over the pre-registered points on a picture displayed in front of us.
This method was among the ideas that I looked at 11 years ago when we were at the very early stage of the development. I was first fascinated by an excellent usability that this kind of method provides, but did not pursue at the end of the day because the level of confidentiality, say, security that can be achieved this way would be far too low.
Identity authentication requires both availability and confidentiality to be satisfied at the same time. Satisfying only one of them will not make a secure identity-verifying solution. It is no bad, though, to hear that Microsoft has started to look to pictorial password systems.
G. Product Line-Up: We have so far come up with the line-up of products as follows:
Mnemonic Guard Library for general purpose (in Japanese)
MG for Web-Access (used at NTT communications for 4 years, bilingual)
MG for PC logon (bilingual, though no English manual available yet)
MG for Smartphone logon (used by 3 corporations, in Japanese)
MG for entrance management (used at Tokyo Soft, in Japanese)
MG with password-manager utilities on USB device (in Japanese)
CryptoMnemo (bilingual)
Authority-distributed CryptoMnemo (bilingual)
Onetime Mnemonic Guard (bilingual)
Anonymity Guard (prototype completed, in Japanese)
of which MG for Web-Access has in itself the sort of phishing-repellent effects as RSA Security's SiteKey but more effectively, and Onetime Mnemonic Guard has the function of neutralizing Trojan Horses by entirely making the memory-based identity authentication data onetime/disposable.
Some products (bilingual) are already available in English. The others can be made available for English-speaking markets with the relatively small work of translation and re-packaging.
Incidentally, the development of the above products were financed by US$1.5 million grants from the Japanese government agencies as well as the US$2 million publicly-guaranteed loans.
