#141 From Neurology to Neuromorphic Computing

Subscribe to get the latest

on Wed May 31 2023 17:00:00 GMT-0700 (Pacific Daylight Time)

with Darren W Pulsipher, Pamela Follet,

In this podcast episode of Embracing Digital Transformation, Dr. Pamela Follett, a neurologist and co-founder of Lewis Rhodes Labs, shares her background and expertise in the field of neurology, specifically with regards to research on the developing brain in early childhood.


#neurology #neuromorphic #security #ai #cybersecurity

Listen Here

As a child neurologist, Dr. Follett emphasizes the importance of understanding the brain’s developmental processes and its remarkable ability to overcome even catastrophic injuries. She shares how her extensive research involved studying rat models and cell cultures to gain insight into children’s development and search for ways to help them get the best outcomes, despite injuries or illnesses. Listening to Dr. Follett’s insights and experiences allows for a better understanding of how neurology is critical in understanding the brain’s processes and how we can better appreciate and support brain development, particularly in children. Dr. Follett also shares her unexpected appearance at a high-tech show, where she utilized her neurology expertise to help a man experiencing seizures during a keynote address. This podcast highlights the potential reach of neurology beyond just medical institutions and the essential role of neurologists in understanding the complexity of the human brain.

Have you ever needed help understanding a complex technology or product, even though you just needed to know how to use it? Dr. has found that her skills in explaining medical problems to patients translate well into explaining complex technologies to non-experts. Her work as a neuroscientist studying brain development led her and her husband to start a high-tech startup that does neuromorphic computing, called Lewis-Rhodes Labs. One of their products, Extreme Search, uses a neuromorphic processor to search through massive amounts of data, mimicking how a brain rapidly recognizes and processes information.

Dr. Follet emphasizes that there are better approaches than mimicking a brain when creating technology. While brains make thousands of mistakes daily, we don’t necessarily want our computers to do the same. Instead, we can take careful lessons from how brains work and apply them to create more efficient and effective technology. This experience highlights the importance of interdisciplinary skills and thinking outside of one’s specific field. By finding skills that transfer across different areas and industries, we can bring unique perspectives and solutions to complex problems.

Extreme search technology is a breakthrough in cyber forensics and real-time analysis that solves the challenge of sifting through Petabytes of unstructured data in record time using cutting-edge hardware and software. Extreme search technology offers high performance, low power consumption, and a constant throughput and is modeled after the human brain designed for high-performance data processing.

Unlike traditional data search methods that require moving data around, extreme search technology allows on-site searches that eliminate all network bottlenecks. The technology suits cyber forensics, cybersecurity, legal discovery, and enterprise data search. Extreme search technology is straightforward for users and requires no new language or pre-identification of patterns but instead uses regular expressions to perform ad-hoc searches of any data described in the text.

Extreme search technology performs sensitive searches on storage appliances and provides real-time analysis identifying potential threats in milliseconds. Combined with traditional detection methods, the technology can detect advanced persistent threats, viruses, adware, trojans, worms, rootkits, and other malware quickly. The use of Extreme Search technology is not limited to cyber forensics. Other research fields, such as genomic research or any unstructured data field, can benefit from Extreme Search technology’s ability to search vast amounts of miscellaneous data in record time.

Many organizations need help finding patterns or insights within their data. Or they have vast databases and struggle to find anything because of the sheer amount of information. Many data scientists resort to indices to decrease the time to find information. This works well when you know what you want when collecting or storing the data. However, many organizations deal with opaque data that fits outside a predetermined structure. In this case, brute-force searches of petabytes of data can take weeks to find common patterns of data not previously determined.

Extreme search technology helps bring visibility to new data areas, allowing for improved analysis and analytics that start with the search and can go faster if data is transformed into the required pieces. This is particularly useful when dealing with healthcare data, where there are massive amounts of structured data, yet information fits outside any database structure.

For more information on Extreme Search technology and approaches to digital transformation, visit Lewis-Rhodes.com.

Podcast Transcript


Hello, this is Darren

Pulsipher, chief solution,architect of public sector at Intel.

And welcome to Embracing

Digital Transformation,where we investigate effective change,leveragingpeople, process and technology.

On today's episode, from neurology toneuromorphic chips for special guest, Dr.

Pamela Follett, neurologistand co-founder of Lewis Rhodes Labs.

Pamela, welcome to the show.

Thank you. It's good to be here.

Hey, Pamela,you and I had a great opportunityto meet at Cyber Tech

Summit two or three weeks ago.

We had a great conversation.

And your background, I just

I was just enamored with your backgroundand your story and everything,and I knew my podcastlisteners would love to hear your story.

So, Pamela,tell us a little bit about yourselfand your background and we'll get well.

We'll talk about it. Okay,

Let's see.

Well, my background

I actually started in engineeringand and then a brief stint inmedical systems, the General Electric.

I walked in a hospital and said, oh,my goodness, I need to work here.

And when to or notto do medical training.

And I went through child mortalityand became a child neurologist.

And then I got very enamoredwith basic neuroscience researchand understanding thethe developing brainin early, really early childhoodpreterm and and around term infants.

And when they had a an injuryand who did well and who didn't.

And I spent a good deal of timestudying that inin the labwith rat models and cell culturesand then and then trying to understandthe children that I saw in the clinic.

So sothat brings up something interesting,because you did the research, but you alsoworked with children at the same time.

So you were still a practicing clinician,right?

With children working through the issuesthat they were dealing with? Yes.

I was at Children's

Hospital in Boston, and I have a there'san entity called Clinician Scientist,which is aphysician, clinical physicianwho also doesbasic ether, basic or clinical research.

But I was doing basic research in the laband then also do clinic.

It was still I see it.

Yeah. Yeah.

This is really fascinating to me becausemy my oldest son has Asperger's syndrome.

So we met with neurologists,we met with psychologists,we met with everybodyto try and figure out what's going on.

Back in the 90.

Yeah, it did.


It and it took some time for themto figure this thing out.

So I've always been fascinatedwith neurology, especially pediatricor child neurology, and I get to talkto a real brain person, right.

Which is different than aa neurosurgeon, Right?

You're very differentthan the neurosurgeon.

Yeah. Surgery.

And I do not do surgery.

So neurosurgeons can, like, fixsome things.

But in nervous system, it's toughto take a child born with spina bifida.

You need a neurosurgeon to repair. The.

Injury to the spinewhere it didn't develop quite right.

But then they're done.

And then you need a neurologist to helpwith the child development and helpthe child get their optimal outcomeout of the body that they ended up with.

But this.

Is a. Really sick.

So a lot of neurology.

Is that becauseneurologists really focus on processand the interactioninside the brain, what's really how thingsare actually really working becauseit's really this nebulous type of thing.

To me, I'm like completely fascinatedby by the whole thing.

But understanding it, I mean, I'm,

I'm talking to a brainiac, obviously.

So the the, the most delightful thing.

Well,there are a lot of delightful things,but the most delightful thingabout being a child neurologistis thatyou don't ever know that you have limitsbecause you don't really knowwhat's going to happen during development.

And so even when there's a catastrophe,you can you can you can hopeand yeah there are somethat it's that that make it really tough.

But it's remarkableeven with the same set of catastrophecircumstance is justhow stunninglywell some children manage to overcomeand it'sbecause their brains aren't static.

It's not just that they're a little adultlearning stuff, okay?

And they have a huge amountto learn to become an adult.

That's not it.

When they're born, their brainisn't there yet.

It's it'snot just a little adult brain, it'sa developing brain.

And andso when something bad happens to it,it can respond in waysthat overcome a lot of the things.

And that's that that has athat has a joy and a hope to itthat that you don't havewith the same injury.

An adult an adult has a massive strokethat takes outhalf of their brainand all of their speech and andand they're devastatedand they might get some of that back.

But, you know, that that that was brainand that brain is now not therewhen a child has that very exactsame strokewhen they're born, which happens the samethe sameblood vessel territory,you can have a stroke in a minuteand a newbornduring that day that it's fora you might not even knowthat the child had as welland some childrendo so well overcoming itwhile the rest of their brainjust overcomes it.

And they do so wellthat some weird thing will happenwhen they have a bike accident at age 14and that somebody will do a scan and say,

Oh my goodness,they had a stroke when they were born.

And that's how well some children do.

And so and that's what you werethat's what you were studying, right?

You were studying.

Why did kids do so welland then some kids didn't.

So you were looking at what what made themovercome these injuries or you come in.

So, yeah, I guess it is a funny word.

It it it's very meaningful in an adult,but it's itwe don't necessarily have it allnailed down exactly what happened.

Sometimes that insult causesa lot of inflammation.

Oh, is that an injury?

Well, it's the result of an insult,but I don't really know what to call that.

So, yeah, I text, I call,

I call all things that that look likeseizure spells because I don't even knowwhat they are until I see them.

And Iknow and,and so but they're all spells, right.

And all the things that hurtthe brain are insult because

I'm not pre deciding what, what happened.


Yeah yeah it's very, very area.

I love the terms.

I'm going to start using them of my wife'sgoing to it's going to drive her crazybut that, that's one of the reasonswhy I'll start using them.

So I Pamela I met you at a show.

First off, why is it neurons areneurologists at a high tech show?

This didn't make any sense to mewhen I first talked to you.

You didn't didn't make any sense to meeither.



We have a medical emergency early on,and I met an EMT over a overa gentleman having a seizure,and he saw me at Worldwide Technologiesbooth and said, What do you do for work?

But, you know, it's.

But you actuallysee, you actually saved this guy, right?

Or you helped himget through a seizure, right? Yeah.

They for a gentlemanhad a seizure during the keynote addressand andour friend from Intel knew I wasthere and called me over and yeah,we were able to make make him feel better,at least safely.

And it was very entertaining for mebecause normally when I go to conferenceis there are 200 physicians thereand the last thing I would want to do itvolunteer to do anything,because there's a whole lot of peopleprobably more qualified than mebecause it'susually not childrenthat are at the conference.

But right.


So I was I was the best in town.

Well, that's. Kind of that's kind of cool.

So is that why youis that why you're trollingall of these tech conferencesso that you can help people out?

I have an. Emergency, you.


So why so why.

Were you there in the first place there?

Because I.

I put on a different hatand I play a neuroscientistfor a high tech startupthat does neuromorphic computing.

But at the moment, there wasn'ta lot of neuroscience to dobecause we have a product andwhen you work with a high tech startup,you do whatever job needs to be done.

And so I have found thatthere are a lot of skillsthat one has from,say, talking to patientswho have a medical problemthat they need to understand.

But they didn't actually do four yearsin medical school and five years inresidency, and maybe they shouldn'thave to have to take care of their child.

So I need to be able to explain to themwhat's wrong and we need to be a team.

And I deeply believe that.

And I've always tried to do thatas well as I could.

But I find that that's not a bad skillwhen you're trying to explaina difficult technology to peoplewho really just need to be able to use it.

And they really don't carehow the processor works.

And so it's kind of fun to find skillsthat you have in one space,show up as useful in another spacewell before you.

Made a big you just said you just said,

Hey, I just switched over it.

No, no, no.

There's a whole story around that.

From the research.

They had the research that you didin understandingbrain developmentor neurological development.

And and when there's insultto that early brain and the onesthat succeeded, that research that you.

Did led. Into the startup of this company.

Is that right? That's right.

That's what happened.

I am I was

I may have mentioned you was working onan animal model and a culture model,and I was trying to insult mymy littlemy little rats and so that I couldunderstand this brain recovery andand I was complaining at homebecause one doesthat one one works well and has to go homeand complain about.

Well, that's it.

That's where we we decompress, right.

With our spouse complaining aboutthe limitations that I hadwith these model systemsand how difficult it wasto see the developmental changefrom one thing I couldput my little ratsin difficult circumstances,but I could give them low oxygen,for example,and then lots and lots of thingshappened, right?

And the whole system gets in trouble.

And rats are remarkable thingsand they recover really well.

But this not necessarily meananything to what children are doing.

And and I couldn't change just one thing.

And I was complaining about thisbecause I wanted to understandthe little subtletiesthat were the differencebetween success and failurefrom a brain recovery standpoint. Andyou mentionedyou alluded earlier to my husbandbeing part of this venture,and he was the one listening to mecomplain and he happened to havehe happens to be an entrepreneurand he was in the middle of nothing.

And was bored.

And he said, Ha,how about a computer model?

And so he developed a simulator for methat of developing neuronswhere you set up a field of neuronsand then you could give them parametersand they would interconnectand then I could change one thingand then they wouldinterconnect a little differently,and then I could do itbriefly and then over time,see what happened and talk changes.

And we developed this model andit was really fun and really interesting.

And I've learned a lot about thingsthat were criticalin, in, in this,like really specific thingsthat were critical.

And and we took that informationor he did.

He said, huh?

As I kept explaining to him what I waslearning and I kept explaining it and why

I thought what he was seeing was wrongbecause I do that a lot. Ieven want to try it.

I tend to do it a lot and,and, and we wrestled with thisa great deal.

And then one day he says, you know,

I think we could makea computer chip out of this.

What you find.

And I said, Go for it.

I'm not interested.

And I went back to my scienceand he called up Sandia National Labs,who had worked with previously and said,

Would you be interestedin a neuromorphic processorthat looked at life like this?

And they were interestedand they thought that was an applicationin cyber securityand we were off and running?

Well, often crawling and often walking.

And and eventually we got running.


To typical startup life, right?

Actually we ran backwards quite a bit.

Yeah. Yes.

That sounds familiar to you.

So you took the fundamental researchyou were doing to help childrenor to understand childrenand maybe even help themif they did have a neonatalor developing brain injury.

And you'veyou've created this incredible technologythat simulates the brain.

A neuromorphic chip.

Yeah, right.


Cool. That's pretty cool.

That's that's not just starting a startmoving to a startupnow that this is pretty goodpretty awesome stuff.

It's been a ride.


Now let's explain a little bitabout what your the product doesthat you did was Sandiaand it's called Extreme Search.

A really interesting conceptthat you guys have here.


So, so when we started,we started with the ideasthat people had brought to computersfrom the brain.

And, and I pointed to theto the way people started to flywhen when the early planesthere was a belief that the wings.

To flap our wings. Yeah. Yeah, yeah.

And because I need to talkas much like a bird as possiblein order to get the ideathat we could get a human up in it.


Well,know that that turned out to be wrong.

What we need to do is understandflight better and then takevery careful things from flight.


And and then stationary wingsand, you know,and then all of the things that have madeflightbetter have not looked more like birds.


Yeah, that's true.

And when Iwhen we started with the processorand people would say, well, talkabout how it needs to look like a brain.

So, you know, if we have a computerthat looks like a brain,it's going to be a computerthat makes thousands of mistakes a day,because that's what brains do.

Brains recognize and iterate andand think and make mistakes and correctand make mistakes.

And and if you think abouthow you see a shadow in the cornerof your bedroom at nightand and you cansee and convince yourself of somethingcompletely different than what's thereso easily when you recognize a personacross the room and you'reyou're not seeing that person,you're always doing this with your brain.

That's but we don't reallywant our computer to do that.

You know, We don't.

Know when I hit my,you know, to create on my mind.

I don't want the computerto think about which letters I meant.

You know, when they're in, when my phonedoes that, it gets pretty annoying.

Oh. Yes, it does.

There are places for this, But.

But we found that in particularlyin cyber security,which is what we're looking at,you don't want mistakes.

You don't want any mistakes. Right.

So we strippedout all of the pieces of our

DAO, all of the pieces of ourof our chip down to the thethe critical featuresthat allowed of the brain that allowed itto function the way we need.

And the pieces that we hung on towere the extra high performanceat very low powerin a constant throughput manner.

So I can walk you through thata little bit.

So you can think really, really hardand your brain doesn'tcatch on fire, right?

Yeah, Yeah, that's true.

They burn a bit more energy,but it's really fairly negligible.

Your brain can work on very, very highperformance on a good day.


But, but it doesn't generate is a real capon the heat that it generatesand it never goes above them.

All right.

It also it also works in real time, right?

You don't you don't bufferwhat you're seeing and what you're hearingand process it laterlike your computer tellsyou do it not you do what you doand you might miss something.

But but it's a real time.

So, you know, when youwhen you have a ten secondspan of time, you have processedall of the input and that 10 seconds,you don't have a choice.


And you do all of thisagain at extremely low power.

So you have a constant throughput,high performance and low power.

Those are the features of the brainthat we stole.

And anotherand by eliminatingall of the learning is mistakes.

And so we couldn't afford any mistakes.

So we got we just we put that asidefor smarter people than us and,and just stuck to the thingsthat that were really neededfor the processor that we were after.

And now we developed a processorthat could just go through datablazingly fast and get to the other sideand tell you what was there.

So you need to findsomethingthat can be in your petabyte of dataand 12 minutes laterwe'll tell you that was it and that's it.

That's the only thing we do.

We do one thing and we do it very well.

So that'swhat I really like, that approach,because that's hard for a startup.

I know because I've done threeand you say at the beginning,

I'm going to be focused,but then you're like, Ooh,

I can make some money over thereand I'm not making money yet,so I need to, I need to make some money.

So I love thatyou guys have been so focused on thisextreme search whichwhich you kind of alluded to a little bit,which is I can find anything in my dataand we're not talking structured data.

This is unstructured data.

I can find anything in that unstructureddata.

A petabyte in 12 minutes, right.

As long as itas long as it's on a on a serverthat that has theextreme search software and hardwareyour IT searches your SSD storagelocally right there in the storageso you don't have to be moving data aroundbecause a lot of the other problemsif you alludedto the structure and things and structurein data is obviously a problem.

Anyone who doesit knows it's it's, it's time consumingand it takes a lot of storagejust to hold your indexes, etc..


But that's only one of the problems.

The other problem isyou have to move the data aroundby having a really low power,high performance processor.

We can put it right next to the startso the data doesn't have to move.

And when the data doesn't have to move,all of the network bottlenecks go away.

All of the searching from a distancegoes away.

All of the all your data is at the edgeproblem, that's fine.

You can just look at it wherever you are.

Where it wherever it goes.

As long as it's sitting in that storage,we can tell youif you need to worry about it.

So what you guys have delivered isthis is a storage appliance.

Yeah, that is that is searchable.

It searches itself.

It searches itself,which is pretty incredible because today,if I need to search large like log filesand that's why you guys went aftercyber security first was because they dealwith a whole bunch of textual logfiles, terabytesand terabytes, up to petabytes of dataand and what happens today is Iwhat do I do?

I use something like Sparkto search for things.

I use Elasticsearch to search for things.


You try to solvethe indexing problem was the factthat you can't reallyindex things you don't knowbeforehand what you're going to want.

Yeah. So? So there you go.

I think that's key.

You guys can do ad hoc searches.

These are not pre these are not prepatterns that you've already identified.

These are like ad hocsearches of patterns in your data.

Anything you can describewith a regular expression.

That's that's incredible.

And it's not a it's not a, it's not a aa whole new language to learn.

It's to Python to come in.

It's, it's not you know, there's no,there's no coursesfor, for users or anything else.

And it's very simple, straightforwardto log in to.

I'm coming along.

So so what would people useuse this technology for?

We mentioned cybersecurity, but what incybersecurity would I use this for?

I'm not using this to do detection, am I?

Well, so that's adicey question because, you know, peoplein cybersecurity, they don't talk much.

No, they don't.

Yeah, yeah, yeah, I've noticed that.

But what we can glean,we figured that cyber forensicwas going to be our sweet spot becausewe kind of originally introduceda product inright after summer winds attack.

And and when people were realizingthat they had been infectedfor eight or nine months or a yearand a half or two,they were sifting throughand they couldn't find the they justcouldn't find it.

So, so much data,pulling it out of cold storage and search.

And then it would take,you know, the iterations, literallytaking weeks, monthsto get through, to find the beginning.

Right. To find the extent. Right.

So we thought, okay,we are sweet in cyber forensics.

You just you know, you can have achunk of storagedepending on how much data you needand you just dump it on there.

You search it as much as you need to.

You can use it,

You know, you can recover you.

And we figured that cyber forensicswas actually thought andand we were talking about it that wayand and we weren't wrong.

I mean, that is But then we have foundour customers approach.

We were told over and over again, Oh, no,that's not what we're using you for.

You're much more valuable in real timeas a as a as a hot bufferwhere we we ingest,but we just dump everything ininto our extreme search boxesand have many we,you know, a handful weeks of everything.

And then when we get caught, which ishappening, you know, daily, these days,multiple timesdaily, we can just we can check,we can make sure we can search.

We can find what filewe have to worry about in this mess.

Also, it's not deep forensics.

It's like real timeforensics. It's so. So.

And the longer the devices have been,devices have been in use.

The So, I mean,this people that won't talk to us, that isto think they're only smart.

You know, the longer they havetheir hands on on a tool,the more uses they foundand then they get really excitednot telling us things.

So I know you guys.

Could you see thisbeing used in other type of like research?

I mean, you're a researcher, so I was justthinking it just popped into my head.

I could see where this couldbe used in genomic research.

This could be used.

I mean, if things are stored as text,

I can searchfor all different types of patternsand yeah,and even even maybe even drug interactionand a whole bunch of different thingsas long as I have things because you guys,you're not doing search on imagesor you're you're searching on.

That's so we have to few other data.

So if you have so many massivenumbers of images and they've been taggedbut you still can't find anythingbecause there's so many of them.

Yeah, there's an embedded data,so there's a space there.

You know, there's an interesting questionbetween the structure and structure.

So you asked about research and lookand it'sit really comes down to the type of datathat people are drowning in.

If if you haveno matter how big it is,if you have a structured databaseand you have indexes foreverything that you need to know,we have very little to add to that.

Yeah, but. I think I find it fascinating.

A lot of timeswe don't know what we don't know.

That is where I was going. You okay? Good.

Yeah, exactly.

So let's say that youthat you've collectedall the electronic medical recordsand there's a lot of thisjust because that's theone side of things that I've seenand that's a lot of structured databecause when you add it into the database,you're up.

But in there is also a whole lot of thingsthat really don't fitwithin any sort of structure.

So as long as you only needwhat you knew you would be looking for,then you're fine.

But what if you want to go backinto your database of anonymized recordsand look for a pattern that you'restarting to wonder might be there?

It's not necessarilygoing to be in your structured data.

And this isn't just true in health care.

This is going to be true in everythingfrom sensor data to oil plots.

Oh, yeah, yeah. I can see this.

Right anywhere where you cansurviving onon the ability to structure your data.

And even if theamount of it isn't overwhelming that yetyou don't you have a lack of resolutionbecause anything you didn't acknowledgeahead of time, you have to search throughway too much datato find it in a reasonable amount of time.

So you don't add labels and you and you'retrying to do machine learning,but you're limited to whatyou already knew.


And so and so, yeah, the ability to dumpdata that that that's opaqueit and and search it is a clear fitbut then the ability to reassessand re index datathat you have half black spot.

Okay you have you have just opaque areaswhere if you didn't know ahead of timeyou were going to be interested.

The text of your medical,of your of your notes,something that gottranscribed, thingsthat were imaged was not a data,all sorts of things,you know, from from a health standpoint.

But this is true all over.

If you whatever you didn't thinkdidn't know you were going to needcan be sitting in there are invisible.

And and so any timethat's kind of energy problem.

That just perfect for you guys.

And so what I usewhen I use extreme searchin conjunction with structuring my data I.

Yeah, did does that make sense?

Could I use them togetherand say I'm going to

I think this is something andthen I type it in an extreme, search itand I get back tons of data and I'm like,holy cow, I should index now on that.


Is it do you see that working together.


So we would never tell anybody to getrid of what, what they have was working.


And you just add extreme searchand how thehow much of it depends onhow big your data problem is.


But you said at the startand then you can make copies of the dataor you can put the datayou weren't sure about thatyou would ever need,but it's still pretty recent.

As the

I said, data is not like the fine lineit does not improve with. It's noand it gives less and less useful.

So so you can use it beside somethingthat you might want to gain visibility on.

You can use it to bring in new thingsthat you might wantto look at differently,but you put it besideso it will anything that's any analysis,any analyticsthat start with search, which these daysas most of them can go faster,if you can transform your datainto the pieces that you need.

And so we can do that for for a systemthat you already have a software languagesthat you're already using,some of these fabulous programsthat are out there that help peoplemanage their their pain and their loss.

And this is in therethat we don't do any of that.

We just tell you where it is.

Pamela, This is wonderful technology.

Where do people find outmore about extreme searchand what you guys have to offer?

I think it's can with our websitethat's www.lewis.... Actually, it isn't any more, is it?

It's just


All right.


That's where they can get informationon the product.

Pamela as always,

I enjoyed talking to you.

A wonderful podcast thanks.

Thank you very much.

Thank you for having me.

Thank you for listeningto Embracing Digital Transformation today.

If you enjoyed our podcast,give it five stars on your favoritepodcasting site or YouTube channel,you can find out more informationabout embracing digital transformationand embracingdigital.org

Until next time, go out and do something.